codex: fix CI failure on PR #13193

Co-authored-by: Codex <noreply@openai.com>

.github/workflows/bazel.yml

@@ -55,6 +55,14 @@ jobs:
         with:
           node-version-file: codex-rs/node-version.txt
 
+      - name: Install Linux build dependencies
+        if: ${{ runner.os == 'Linux' }}
+        shell: bash
+        run: |
+          set -euo pipefail
+          sudo apt-get update -y
+          sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends libasound2-dev pkg-config libcap-dev
+
       # Some integration tests rely on DotSlash being installed.
       # See https://github.com/openai/codex/pull/7617.
       - name: Install DotSlash

.github/workflows/rust-ci.yml

@@ -190,7 +190,7 @@ jobs:
           set -euo pipefail
           if command -v apt-get >/dev/null 2>&1; then
             sudo apt-get update -y
-            packages=(pkg-config libcap-dev)
+            packages=(libasound2-dev pkg-config libcap-dev)
             if [[ "${{ matrix.target }}" == 'x86_64-unknown-linux-musl' || "${{ matrix.target }}" == 'aarch64-unknown-linux-musl' ]]; then
               packages+=(libubsan1)
             fi
@@ -510,7 +510,7 @@ jobs:
           set -euo pipefail
           if command -v apt-get >/dev/null 2>&1; then
             sudo apt-get update -y
-            sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends pkg-config libcap-dev
+            sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends libasound2-dev pkg-config libcap-dev
           fi
       # Some integration tests rely on DotSlash being installed.
       # See https://github.com/openai/codex/pull/7617.

.github/workflows/rust-release.yml

@@ -107,13 +107,13 @@ jobs:
           echo "Total RAM: ${total_ram}"
           echo "Disk usage:"
           df -h .
-      - name: Install Linux bwrap build dependencies
+      - name: Install Linux build dependencies
         if: ${{ runner.os == 'Linux' }}
         shell: bash
         run: |
           set -euo pipefail
           sudo apt-get update -y
-          sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends pkg-config libcap-dev
+          sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends libasound2-dev pkg-config libcap-dev
       - name: Install UBSan runtime (musl)
         if: ${{ matrix.target == 'x86_64-unknown-linux-musl' || matrix.target == 'aarch64-unknown-linux-musl' }}
         shell: bash

.github/workflows/sdk.yml

@@ -18,7 +18,7 @@ jobs:
         run: |
           set -euo pipefail
           sudo apt-get update -y
-          sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends pkg-config libcap-dev
+          sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends libasound2-dev pkg-config libcap-dev
 
       - name: Setup pnpm
         uses: pnpm/action-setup@v4

README.md

@@ -42,6 +42,8 @@ Each GitHub Release contains many executables, but in practice, you likely want
 
 Each archive contains a single entry with the platform baked into the name (e.g., `codex-x86_64-unknown-linux-musl`), so you likely want to rename it to `codex` after extracting it.
 
+On Linux, these prebuilt binaries are the recommended install path and should work on most mainstream desktops with ALSA available. If you build Codex from source on Linux instead, install the ALSA development headers first (for example, `libasound2-dev` on Debian/Ubuntu).
+
 </details>
 
 ### Using Codex with your ChatGPT plan

codex-rs/Cargo.lock

@@ -293,9 +293,9 @@ checksum = "683d7910e743518b0e34f1186f92494becacb047c7b6bf616c96772180fef923"
 
 [[package]]
 name = "alsa"
-version = "0.9.1"
+version = "0.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ed7572b7ba83a31e20d1b48970ee402d2e3e0537dcfe0a3ff4d6eb7508617d43"
+checksum = "812947049edcd670a82cd5c73c3661d2e58468577ba8489de58e1a73c04cbd5d"
 dependencies = [
  "alsa-sys",
  "bitflags 2.10.0",
@@ -305,9 +305,9 @@ dependencies = [
 
 [[package]]
 name = "alsa-sys"
-version = "0.3.1"
+version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db8fee663d06c4e303404ef5f40488a53e062f89ba8bfed81f42325aafad1527"
+checksum = "ad7569085a265dd3f607ebecce7458eaab2132a84393534c95b18dcbc3f31e04"
 dependencies = [
  "libc",
  "pkg-config",
@@ -380,7 +380,7 @@ version = "1.1.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "40c48f72fd53cd289104fc64099abca73db4166ad86ea0b4341abe65af83dadc"
 dependencies = [
- "windows-sys 0.61.2",
+ "windows-sys 0.60.2",
 ]
 
 [[package]]
@@ -391,7 +391,7 @@ checksum = "291e6a250ff86cd4a820112fb8898808a366d8f9f58ce16d1f538353ad55747d"
 dependencies = [
  "anstyle",
  "once_cell_polyfill",
- "windows-sys 0.61.2",
+ "windows-sys 0.60.2",
 ]
 
 [[package]]
@@ -935,24 +935,6 @@ dependencies = [
  "serde",
 ]
 
-[[package]]
-name = "bindgen"
-version = "0.72.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "993776b509cfb49c750f11b8f07a46fa23e0a1386ffc01fb1e7d343efc387895"
-dependencies = [
- "bitflags 2.10.0",
- "cexpr",
- "clang-sys",
- "itertools 0.13.0",
- "proc-macro2",
- "quote",
- "regex",
- "rustc-hash 2.1.1",
- "shlex",
- "syn 2.0.114",
-]
-
 [[package]]
 name = "bit-set"
 version = "0.5.3"
@@ -1190,15 +1172,6 @@ version = "1.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6d43a04d8753f35258c91f8ec639f792891f748a1edbd759cf1dcea3382ad83c"
 
-[[package]]
-name = "cexpr"
-version = "0.6.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6fac387a98bb7c37292057cffc56d62ecb629900026402633ae9160df93a8766"
-dependencies = [
- "nom 7.1.3",
-]
-
 [[package]]
 name = "cfg-if"
 version = "1.0.4"
@@ -1283,17 +1256,6 @@ dependencies = [
  "zeroize",
 ]
 
-[[package]]
-name = "clang-sys"
-version = "1.8.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0b023947811758c97c59bf9d1c188fd619ad4718dcaa767947df1cadb14f39f4"
-dependencies = [
- "glob",
- "libc",
- "libloading",
-]
-
 [[package]]
 name = "clap"
 version = "4.5.58"
@@ -2856,32 +2818,25 @@ dependencies = [
 
 [[package]]
 name = "coreaudio-rs"
-version = "0.11.3"
+version = "0.14.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "321077172d79c662f64f5071a03120748d5bb652f5231570141be24cfcd2bace"
+checksum = "d15c3c3cee7c087938f7ad1c3098840b3ef1f1bdc7f6e496336c3b1e7a6f3914"
 dependencies = [
- "bitflags 1.3.2",
- "core-foundation-sys",
- "coreaudio-sys",
-]
-
-[[package]]
-name = "coreaudio-sys"
-version = "0.2.17"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ceec7a6067e62d6f931a2baf6f3a751f4a892595bcec1461a3c94ef9949864b6"
-dependencies = [
- "bindgen",
+ "bitflags 2.10.0",
+ "libc",
+ "objc2-audio-toolbox",
+ "objc2-core-audio",
+ "objc2-core-audio-types",
+ "objc2-core-foundation",
 ]
 
 [[package]]
 name = "cpal"
-version = "0.15.3"
+version = "0.17.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "873dab07c8f743075e57f524c583985fbaf745602acbe916a01539364369a779"
+checksum = "d8942da362c0f0d895d7cac616263f2f9424edc5687364dfd1d25ef7eba506d7"
 dependencies = [
  "alsa",
- "core-foundation-sys",
  "coreaudio-rs",
  "dasp_sample",
  "jni",
@@ -2890,11 +2845,19 @@ dependencies = [
  "mach2",
  "ndk",
  "ndk-context",
- "oboe",
+ "num-derive",
+ "num-traits",
+ "objc2",
+ "objc2-audio-toolbox",
+ "objc2-avf-audio",
+ "objc2-core-audio",
+ "objc2-core-audio-types",
+ "objc2-core-foundation",
+ "objc2-foundation",
  "wasm-bindgen",
  "wasm-bindgen-futures",
  "web-sys",
- "windows 0.54.0",
+ "windows 0.62.2",
 ]
 
 [[package]]
@@ -3452,7 +3415,7 @@ dependencies = [
  "libc",
  "option-ext",
  "redox_users 0.5.2",
- "windows-sys 0.61.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -3697,7 +3660,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb"
 dependencies = [
  "libc",
- "windows-sys 0.61.2",
+ "windows-sys 0.52.0",
 ]
 
 [[package]]
@@ -4228,12 +4191,6 @@ version = "0.32.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "e629b9b98ef3dd8afe6ca2bd0f89306cec16d43d907889945bc5d6687f2f13c7"
 
-[[package]]
-name = "glob"
-version = "0.3.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0cc23270f6e1808e30a928bdc84dea0b9b4136a8bc82338574f23baf47bbd280"
-
 [[package]]
 name = "globset"
 version = "0.4.18"
@@ -4692,7 +4649,7 @@ dependencies = [
  "js-sys",
  "log",
  "wasm-bindgen",
- "windows-core 0.62.2",
+ "windows-core 0.58.0",
 ]
 
 [[package]]
@@ -5090,7 +5047,7 @@ checksum = "3640c1c38b8e4e43584d8df18be5fc6b0aa314ce6ebf51b53313d4306cca8e46"
 dependencies = [
  "hermit-abi",
  "libc",
- "windows-sys 0.61.2",
+ "windows-sys 0.52.0",
 ]
 
 [[package]]
@@ -5314,16 +5271,6 @@ dependencies = [
  "pkg-config",
 ]
 
-[[package]]
-name = "libloading"
-version = "0.8.9"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d7c4b02199fee7c5d21a5ae7d8cfa79a6ef5bb2fc834d6e9058e89c825efdc55"
-dependencies = [
- "cfg-if",
- "windows-link",
-]
-
 [[package]]
 name = "libm"
 version = "0.2.16"
@@ -5517,9 +5464,9 @@ dependencies = [
 
 [[package]]
 name = "mach2"
-version = "0.4.3"
+version = "0.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d640282b302c0bb0a2a8e0233ead9035e3bed871f0b7e81fe4a1ec829765db44"
+checksum = "6a1b95cd5421ec55b445b5ae102f5ea0e768de1f82bd3001e11f426c269c3aea"
 dependencies = [
  "libc",
 ]
@@ -5709,9 +5656,9 @@ dependencies = [
 
 [[package]]
 name = "ndk"
-version = "0.8.0"
+version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2076a31b7010b17a38c01907c45b945e8f11495ee4dd588309718901b1f7a5b7"
+checksum = "c3f42e7bbe13d351b6bead8286a43aac9534b82bd3cc43e47037f012ebfd62d4"
 dependencies = [
  "bitflags 2.10.0",
  "jni-sys",
@@ -5729,9 +5676,9 @@ checksum = "27b02d87554356db9e9a873add8782d4ea6e3e58ea071a9adb9a2e8ddb884a8b"
 
 [[package]]
 name = "ndk-sys"
-version = "0.5.0+25.2.9519653"
+version = "0.6.0+11769913"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8c196769dd60fd4f363e11d948139556a344e79d451aeb2fa2fd040738ef7691"
+checksum = "ee6cda3051665f1fb8d9e08fc35c96d5a244fb1be711a03b71118828afc9a873"
 dependencies = [
  "jni-sys",
 ]
@@ -5846,7 +5793,7 @@ version = "0.50.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7957b9740744892f114936ab4a57b3f487491bbeafaf8083688b16841a4240e5"
 dependencies = [
- "windows-sys 0.61.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -6057,6 +6004,31 @@ dependencies = [
  "objc2-foundation",
 ]
 
+[[package]]
+name = "objc2-audio-toolbox"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6948501a91121d6399b79abaa33a8aa4ea7857fe019f341b8c23ad6e81b79b08"
+dependencies = [
+ "bitflags 2.10.0",
+ "libc",
+ "objc2",
+ "objc2-core-audio",
+ "objc2-core-audio-types",
+ "objc2-core-foundation",
+ "objc2-foundation",
+]
+
+[[package]]
+name = "objc2-avf-audio"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "13a380031deed8e99db00065c45937da434ca987c034e13b87e4441f9e4090be"
+dependencies = [
+ "objc2",
+ "objc2-foundation",
+]
+
 [[package]]
 name = "objc2-cloud-kit"
 version = "0.3.2"
@@ -6068,6 +6040,29 @@ dependencies = [
  "objc2-foundation",
 ]
 
+[[package]]
+name = "objc2-core-audio"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e1eebcea8b0dbff5f7c8504f3107c68fc061a3eb44932051c8cf8a68d969c3b2"
+dependencies = [
+ "dispatch2",
+ "objc2",
+ "objc2-core-audio-types",
+ "objc2-core-foundation",
+ "objc2-foundation",
+]
+
+[[package]]
+name = "objc2-core-audio-types"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5a89f2ec274a0cf4a32642b2991e8b351a404d290da87bb6a9a9d8632490bd1c"
+dependencies = [
+ "bitflags 2.10.0",
+ "objc2",
+]
+
 [[package]]
 name = "objc2-core-data"
 version = "0.3.2"
@@ -6085,7 +6080,9 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2a180dd8642fa45cdb7dd721cd4c11b1cadd4929ce112ebd8b9f5803cc79d536"
 dependencies = [
  "bitflags 2.10.0",
+ "block2",
  "dispatch2",
+ "libc",
  "objc2",
 ]
 
@@ -6216,29 +6213,6 @@ dependencies = [
  "memchr",
 ]
 
-[[package]]
-name = "oboe"
-version = "0.6.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e8b61bebd49e5d43f5f8cc7ee2891c16e0f41ec7954d36bcb6c14c5e0de867fb"
-dependencies = [
- "jni",
- "ndk",
- "ndk-context",
- "num-derive",
- "num-traits",
- "oboe-sys",
-]
-
-[[package]]
-name = "oboe-sys"
-version = "0.6.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6c8bb09a4a2b1d668170cfe0a7d5bc103f8999fb316c98099b6a9939c9f2e79d"
-dependencies = [
- "cc",
-]
-
 [[package]]
 name = "oid-registry"
 version = "0.8.1"
@@ -6490,7 +6464,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7d8fae84b431384b68627d0f9b3b1245fcf9f46f6c0e3dc902e9dce64edd1967"
 dependencies = [
  "libc",
- "windows-sys 0.61.2",
+ "windows-sys 0.48.0",
 ]
 
 [[package]]
@@ -7891,7 +7865,7 @@ dependencies = [
  "errno",
  "libc",
  "linux-raw-sys 0.11.0",
- "windows-sys 0.61.2",
+ "windows-sys 0.52.0",
 ]
 
 [[package]]
@@ -9289,7 +9263,7 @@ dependencies = [
  "getrandom 0.3.4",
  "once_cell",
  "rustix 1.1.3",
- "windows-sys 0.61.2",
+ "windows-sys 0.52.0",
 ]
 
 [[package]]
@@ -10661,7 +10635,7 @@ version = "0.1.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22"
 dependencies = [
- "windows-sys 0.61.2",
+ "windows-sys 0.48.0",
 ]
 
 [[package]]
@@ -10670,16 +10644,6 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
 
-[[package]]
-name = "windows"
-version = "0.54.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9252e5725dbed82865af151df558e754e4a3c2c30818359eb17465f1346a1b49"
-dependencies = [
- "windows-core 0.54.0",
- "windows-targets 0.52.6",
-]
-
 [[package]]
 name = "windows"
 version = "0.58.0"
@@ -10711,16 +10675,6 @@ dependencies = [
  "windows-core 0.62.2",
 ]
 
-[[package]]
-name = "windows-core"
-version = "0.54.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "12661b9c89351d684a50a8a643ce5f608e20243b9fb84687800163429f161d65"
-dependencies = [
- "windows-result 0.1.2",
- "windows-targets 0.52.6",
-]
-
 [[package]]
 name = "windows-core"
 version = "0.58.0"
@@ -10829,15 +10783,6 @@ dependencies = [
  "windows-strings 0.5.1",
 ]
 
-[[package]]
-name = "windows-result"
-version = "0.1.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5e383302e8ec8515204254685643de10811af0ed97ea37210dc26fb0032647f8"
-dependencies = [
- "windows-targets 0.52.6",
-]
-
 [[package]]
 name = "windows-result"
 version = "0.2.0"

codex-rs/README.md

@@ -1,6 +1,6 @@
 # Codex CLI (Rust Implementation)
 
-We provide Codex CLI as a standalone, native executable to ensure a zero-dependency install.
+We provide Codex CLI as a standalone, native executable. For most users, the simplest install path is a prebuilt package or release binary.
 
 ## Installing Codex
 
@@ -13,6 +13,8 @@ codex
 
 You can also install via Homebrew (`brew install --cask codex`) or download a platform-specific release directly from our [GitHub Releases](https://github.com/openai/codex/releases).
 
+On Linux, the prebuilt release binaries are the recommended path. If you build from source on Linux, install ALSA development headers first (for example, `libasound2-dev` on Debian/Ubuntu).
+
 ## Documentation quickstart
 
 - First run with Codex? Start with [`docs/getting-started.md`](../docs/getting-started.md) (links to the walkthrough for prompts, keyboard shortcuts, and session management).

codex-rs/core/config.schema.json

@@ -1588,7 +1588,7 @@
         }
       ],
       "default": null,
-      "description": "Machine-local realtime audio device preferences used by realtime voice."
+      "description": "Machine-local audio device preferences. `audio.microphone` applies to transcription and realtime voice input. `audio.speaker` applies to realtime voice output."
     },
     "background_terminal_max_timeout": {
       "description": "Maximum poll window for background terminal output (`write_stdin`), in milliseconds. Default: `300000` (5 minutes).",

codex-rs/core/src/config/edit.rs

@@ -872,7 +872,7 @@ impl ConfigEditsBuilder {
         self
     }
 
-    pub fn set_realtime_microphone(mut self, microphone: Option<&str>) -> Self {
+    pub fn set_audio_microphone(mut self, microphone: Option<&str>) -> Self {
         let segments = vec!["audio".to_string(), "microphone".to_string()];
         match microphone {
             Some(microphone) => self.edits.push(ConfigEdit::SetPath {
@@ -884,7 +884,7 @@ impl ConfigEditsBuilder {
         self
     }
 
-    pub fn set_realtime_speaker(mut self, speaker: Option<&str>) -> Self {
+    pub fn set_audio_speaker(mut self, speaker: Option<&str>) -> Self {
         let segments = vec!["audio".to_string(), "speaker".to_string()];
         match speaker {
             Some(speaker) => self.edits.push(ConfigEdit::SetPath {
@@ -1893,8 +1893,8 @@ model_reasoning_effort = "high"
         let codex_home = tmp.path();
 
         ConfigEditsBuilder::new(codex_home)
-            .set_realtime_microphone(Some("USB Mic"))
-            .set_realtime_speaker(Some("Desk Speakers"))
+            .set_audio_microphone(Some("USB Mic"))
+            .set_audio_speaker(Some("Desk Speakers"))
             .apply_blocking()
             .expect("persist realtime audio");
 
@@ -1914,7 +1914,7 @@ model_reasoning_effort = "high"
         );
 
         ConfigEditsBuilder::new(codex_home)
-            .set_realtime_microphone(None)
+            .set_audio_microphone(None)
             .apply_blocking()
             .expect("clear realtime microphone");
 

codex-rs/core/src/config/mod.rs

@@ -1182,7 +1182,9 @@ pub struct ConfigToml {
     /// Base URL for requests to ChatGPT (as opposed to the OpenAI API).
     pub chatgpt_base_url: Option<String>,
 
-    /// Machine-local realtime audio device preferences used by realtime voice.
+    /// Machine-local audio device preferences.
+    /// `audio.microphone` applies to transcription and realtime voice input.
+    /// `audio.speaker` applies to realtime voice output.
     #[serde(default)]
     pub audio: Option<RealtimeAudioToml>,
 
@@ -6132,8 +6134,8 @@ experimental_realtime_ws_backend_prompt = "prompt from config"
         let cfg: ConfigToml = toml::from_str(
             r#"
 [audio]
-microphone = "USB Mic"
-speaker = "Desk Speakers"
+microphone = "input-device-id"
+speaker = "output-device-id"
 "#,
         )
         .expect("TOML deserialization should succeed");
@@ -6142,8 +6144,11 @@ speaker = "Desk Speakers"
             .audio
             .as_ref()
             .expect("realtime audio config should be present");
-        assert_eq!(realtime_audio.microphone.as_deref(), Some("USB Mic"));
-        assert_eq!(realtime_audio.speaker.as_deref(), Some("Desk Speakers"));
+        assert_eq!(
+            realtime_audio.microphone.as_deref(),
+            Some("input-device-id")
+        );
+        assert_eq!(realtime_audio.speaker.as_deref(), Some("output-device-id"));
 
         let codex_home = TempDir::new()?;
         let config = Config::load_from_base_config_with_overrides(
@@ -6152,10 +6157,13 @@ speaker = "Desk Speakers"
             codex_home.path().to_path_buf(),
         )?;
 
-        assert_eq!(config.realtime_audio.microphone.as_deref(), Some("USB Mic"));
+        assert_eq!(
+            config.realtime_audio.microphone.as_deref(),
+            Some("input-device-id")
+        );
         assert_eq!(
             config.realtime_audio.speaker.as_deref(),
-            Some("Desk Speakers")
+            Some("output-device-id")
         );
         Ok(())
     }

codex-rs/tui/Cargo.toml

@@ -13,12 +13,11 @@ name = "codex_tui"
 path = "src/lib.rs"
 
 [features]
-default = ["voice-input"]
+default = []
 # Enable vt100-based tests (emulator) when running with `--features vt100-tests`.
 vt100-tests = []
 # Gate verbose debug logging inside the TUI implementation.
 debug-logs = []
-voice-input = ["dep:cpal", "dep:hound"]
 
 [lints]
 workspace = true
@@ -103,14 +102,12 @@ unicode-width = { workspace = true }
 url = { workspace = true }
 webbrowser = { workspace = true }
 uuid = { workspace = true }
+cpal = { version = "0.17.3" }
+hound = { version = "3.5" }
 
 codex-windows-sandbox = { workspace = true }
 tokio-util = { workspace = true, features = ["time"] }
 
-[target.'cfg(not(target_os = "linux"))'.dependencies]
-cpal = { version = "0.15", optional = true }
-hound = { version = "3.5", optional = true }
-
 [target.'cfg(unix)'.dependencies]
 libc = { workspace = true }
 

codex-rs/tui/src/app.rs

@@ -1,7 +1,7 @@
 use crate::app_backtrack::BacktrackState;
 use crate::app_event::AppEvent;
 use crate::app_event::ExitMode;
-use crate::app_event::RealtimeAudioDeviceKind;
+use crate::app_event::VoiceAudioDeviceKind;
 #[cfg(target_os = "windows")]
 use crate::app_event::WindowsSandboxEnableMode;
 use crate::app_event_sender::AppEventSender;
@@ -2228,8 +2228,8 @@ impl App {
             AppEvent::UpdatePersonality(personality) => {
                 self.on_update_personality(personality);
             }
-            AppEvent::OpenRealtimeAudioDeviceSelection { kind } => {
-                self.chat_widget.open_realtime_audio_device_selection(kind);
+            AppEvent::OpenVoiceAudioDeviceSelection { kind } => {
+                self.chat_widget.open_voice_audio_device_selection(kind);
             }
             AppEvent::OpenReasoningPopup { model } => {
                 self.chat_widget.open_reasoning_popup(model);
@@ -2656,37 +2656,38 @@ impl App {
                     }
                 }
             }
-            AppEvent::PersistRealtimeAudioDeviceSelection { kind, name } => {
+            AppEvent::PersistVoiceAudioDeviceSelection { kind, device_id } => {
                 let builder = match kind {
-                    RealtimeAudioDeviceKind::Microphone => {
+                    VoiceAudioDeviceKind::Microphone => {
                         ConfigEditsBuilder::new(&self.config.codex_home)
-                            .set_realtime_microphone(name.as_deref())
+                            .set_audio_microphone(device_id.as_deref())
                     }
-                    RealtimeAudioDeviceKind::Speaker => {
+                    VoiceAudioDeviceKind::Speaker => {
                         ConfigEditsBuilder::new(&self.config.codex_home)
-                            .set_realtime_speaker(name.as_deref())
+                            .set_audio_speaker(device_id.as_deref())
                     }
                 };
 
                 match builder.apply().await {
                     Ok(()) => {
                         match kind {
-                            RealtimeAudioDeviceKind::Microphone => {
-                                self.config.realtime_audio.microphone = name.clone();
+                            VoiceAudioDeviceKind::Microphone => {
+                                self.config.realtime_audio.microphone = device_id.clone();
                             }
-                            RealtimeAudioDeviceKind::Speaker => {
-                                self.config.realtime_audio.speaker = name.clone();
+                            VoiceAudioDeviceKind::Speaker => {
+                                self.config.realtime_audio.speaker = device_id.clone();
                             }
                         }
                         self.chat_widget
-                            .set_realtime_audio_device(kind, name.clone());
+                            .set_voice_audio_device(kind, device_id.clone());
 
                         if self.chat_widget.realtime_conversation_is_live() {
-                            self.chat_widget.open_realtime_audio_restart_prompt(kind);
+                            self.chat_widget.open_voice_audio_restart_prompt(kind);
                         } else {
-                            let selection = name.unwrap_or_else(|| "System default".to_string());
+                            let selection =
+                                device_id.unwrap_or_else(|| "System default".to_string());
                             self.chat_widget.add_info_message(
-                                format!("Realtime {} set to {selection}", kind.noun()),
+                                format!("Audio {} set to {selection}", kind.noun()),
                                 None,
                             );
                         }
@@ -2694,17 +2695,15 @@ impl App {
                     Err(err) => {
                         tracing::error!(
                             error = %err,
-                            "failed to persist realtime audio selection"
+                            "failed to persist audio device selection"
                         );
-                        self.chat_widget.add_error_message(format!(
-                            "Failed to save realtime {}: {err}",
-                            kind.noun()
-                        ));
+                        self.chat_widget
+                            .add_error_message(format!("Failed to save {}: {err}", kind.noun()));
                     }
                 }
             }
-            AppEvent::RestartRealtimeAudioDevice { kind } => {
-                self.chat_widget.restart_realtime_audio_device(kind);
+            AppEvent::RestartVoiceAudioDevice { kind } => {
+                self.chat_widget.restart_voice_audio_device(kind);
             }
             AppEvent::UpdateAskForApprovalPolicy(policy) => {
                 self.runtime_approval_policy_override = Some(policy);
@@ -3099,15 +3098,13 @@ impl App {
                     ));
                 }
             },
-            #[cfg(not(target_os = "linux"))]
             AppEvent::TranscriptionComplete { id, text } => {
                 self.chat_widget.replace_transcription(&id, &text);
             }
-            #[cfg(not(target_os = "linux"))]
-            AppEvent::TranscriptionFailed { id, error: _ } => {
+            AppEvent::TranscriptionFailed { id, error } => {
+                tracing::error!("voice transcription failed: {error}");
                 self.chat_widget.remove_transcription_placeholder(&id);
             }
-            #[cfg(not(target_os = "linux"))]
             AppEvent::UpdateRecordingMeter { id, text } => {
                 // Update in place to preserve the element id for subsequent frames.
                 let updated = self.chat_widget.update_transcription_in_place(&id, &text);

codex-rs/tui/src/app_event.rs

@@ -30,12 +30,12 @@ use codex_protocol::protocol::AskForApproval;
 use codex_protocol::protocol::SandboxPolicy;
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
-pub(crate) enum RealtimeAudioDeviceKind {
+pub(crate) enum VoiceAudioDeviceKind {
     Microphone,
     Speaker,
 }
 
-impl RealtimeAudioDeviceKind {
+impl VoiceAudioDeviceKind {
     pub(crate) fn title(self) -> &'static str {
         match self {
             Self::Microphone => "Microphone",
@@ -198,23 +198,19 @@ pub(crate) enum AppEvent {
     },
 
     /// Open the device picker for a realtime microphone or speaker.
-    OpenRealtimeAudioDeviceSelection {
-        kind: RealtimeAudioDeviceKind,
+    OpenVoiceAudioDeviceSelection {
+        kind: VoiceAudioDeviceKind,
     },
 
-    /// Persist the selected realtime microphone or speaker to top-level config.
-    #[cfg_attr(
-        any(target_os = "linux", not(feature = "voice-input")),
-        allow(dead_code)
-    )]
-    PersistRealtimeAudioDeviceSelection {
-        kind: RealtimeAudioDeviceKind,
-        name: Option<String>,
+    /// Persist the selected microphone or speaker device id to top-level config.
+    PersistVoiceAudioDeviceSelection {
+        kind: VoiceAudioDeviceKind,
+        device_id: Option<String>,
     },
 
-    /// Restart the selected realtime microphone or speaker locally.
-    RestartRealtimeAudioDevice {
-        kind: RealtimeAudioDeviceKind,
+    /// Restart the selected microphone or speaker locally.
+    RestartVoiceAudioDevice {
+        kind: VoiceAudioDeviceKind,
     },
 
     /// Open the reasoning selection popup after picking a model.
@@ -377,21 +373,18 @@ pub(crate) enum AppEvent {
 
     /// Live update for the in-progress voice recording placeholder. Carries
     /// the placeholder `id` and the text to display (e.g., an ASCII meter).
-    #[cfg(not(target_os = "linux"))]
     UpdateRecordingMeter {
         id: String,
         text: String,
     },
 
     /// Voice transcription finished for the given placeholder id.
-    #[cfg(not(target_os = "linux"))]
     TranscriptionComplete {
         id: String,
         text: String,
     },
 
     /// Voice transcription failed; remove the placeholder identified by `id`.
-    #[cfg(not(target_os = "linux"))]
     TranscriptionFailed {
         id: String,
         #[allow(dead_code)]

codex-rs/tui/src/audio_device.rs

@@ -3,37 +3,53 @@ use cpal::traits::DeviceTrait;
 use cpal::traits::HostTrait;
 use tracing::warn;
 
-use crate::app_event::RealtimeAudioDeviceKind;
+use crate::app_event::VoiceAudioDeviceKind;
 
 const PREFERRED_INPUT_SAMPLE_RATE: u32 = 24_000;
 const PREFERRED_INPUT_CHANNELS: u16 = 1;
 
-pub(crate) fn list_realtime_audio_device_names(
-    kind: RealtimeAudioDeviceKind,
-) -> Result<Vec<String>, String> {
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub(crate) struct VoiceAudioDevice {
+    pub(crate) id: String,
+    pub(crate) label: String,
+}
+
+pub(crate) fn list_voice_audio_devices(
+    kind: VoiceAudioDeviceKind,
+) -> Result<Vec<VoiceAudioDevice>, String> {
     let host = cpal::default_host();
-    let mut device_names = Vec::new();
-    for device in devices(&host, kind)? {
-        let Ok(name) = device.name() else {
-            continue;
-        };
-        if !device_names.contains(&name) {
-            device_names.push(name);
-        }
-    }
-    Ok(device_names)
+    Ok(devices(&host, kind)?
+        .into_iter()
+        .filter_map(|device| {
+            let label = device.description().ok()?.to_string();
+            let id = device.id().ok()?.to_string();
+            Some(VoiceAudioDevice { id, label })
+        })
+        .collect())
 }
 
 pub(crate) fn select_configured_input_device_and_config(
     config: &Config,
 ) -> Result<(cpal::Device, cpal::SupportedStreamConfig), String> {
-    select_device_and_config(RealtimeAudioDeviceKind::Microphone, config)
+    select_device_and_config(
+        VoiceAudioDeviceKind::Microphone,
+        config.realtime_audio.microphone.as_deref(),
+    )
 }
 
 pub(crate) fn select_configured_output_device_and_config(
     config: &Config,
 ) -> Result<(cpal::Device, cpal::SupportedStreamConfig), String> {
-    select_device_and_config(RealtimeAudioDeviceKind::Speaker, config)
+    select_device_and_config(
+        VoiceAudioDeviceKind::Speaker,
+        config.realtime_audio.speaker.as_deref(),
+    )
+}
+
+pub(crate) fn select_input_device_and_config(
+    device_id: Option<&str>,
+) -> Result<(cpal::Device, cpal::SupportedStreamConfig), String> {
+    select_device_and_config(VoiceAudioDeviceKind::Microphone, device_id)
 }
 
 pub(crate) fn preferred_input_config(
@@ -52,7 +68,7 @@ pub(crate) fn preferred_input_config(
                 _ => return None,
             };
             let sample_rate = preferred_input_sample_rate(&range);
-            let sample_rate_penalty = sample_rate.0.abs_diff(PREFERRED_INPUT_SAMPLE_RATE);
+            let sample_rate_penalty = sample_rate.abs_diff(PREFERRED_INPUT_SAMPLE_RATE);
             let channel_penalty = range.channels().abs_diff(PREFERRED_INPUT_CHANNELS);
             Some((
                 (sample_rate_penalty, channel_penalty, sample_format_rank),
@@ -66,111 +82,104 @@ pub(crate) fn preferred_input_config(
 }
 
 fn select_device_and_config(
-    kind: RealtimeAudioDeviceKind,
-    config: &Config,
+    kind: VoiceAudioDeviceKind,
+    configured_device_id: Option<&str>,
 ) -> Result<(cpal::Device, cpal::SupportedStreamConfig), String> {
     let host = cpal::default_host();
-    let configured_name = configured_name(kind, config);
-    let selected = configured_name
-        .and_then(|name| find_device_by_name(&host, kind, name))
+    let selected = configured_device_id
+        .and_then(|device_id| configured_device(&host, kind, device_id))
         .or_else(|| {
             let default_device = default_device(&host, kind);
-            if let Some(name) = configured_name && default_device.is_some() {
+            if let Some(device_id) = configured_device_id && default_device.is_some() {
                 warn!(
-                    "configured {} audio device `{name}` was unavailable; falling back to system default",
+                    "configured {} audio device `{device_id}` was unavailable; falling back to system default",
                     kind.noun()
                 );
             }
             default_device
         })
-        .ok_or_else(|| missing_device_error(kind, configured_name))?;
+        .ok_or_else(|| missing_device_error(kind, configured_device_id))?;
 
     let stream_config = match kind {
-        RealtimeAudioDeviceKind::Microphone => preferred_input_config(&selected)?,
-        RealtimeAudioDeviceKind::Speaker => default_config(&selected, kind)?,
+        VoiceAudioDeviceKind::Microphone => preferred_input_config(&selected)?,
+        VoiceAudioDeviceKind::Speaker => default_config(&selected, kind)?,
     };
     Ok((selected, stream_config))
 }
 
-fn configured_name(kind: RealtimeAudioDeviceKind, config: &Config) -> Option<&str> {
+fn configured_device(
+    host: &cpal::Host,
+    kind: VoiceAudioDeviceKind,
+    device_id: &str,
+) -> Option<cpal::Device> {
+    let parsed_id = device_id.parse().ok()?;
+    let device = host.device_by_id(&parsed_id)?;
+
     match kind {
-        RealtimeAudioDeviceKind::Microphone => config.realtime_audio.microphone.as_deref(),
-        RealtimeAudioDeviceKind::Speaker => config.realtime_audio.speaker.as_deref(),
+        VoiceAudioDeviceKind::Microphone => device.default_input_config().ok().map(|_| device),
+        VoiceAudioDeviceKind::Speaker => device.default_output_config().ok().map(|_| device),
     }
 }
 
-fn find_device_by_name(
-    host: &cpal::Host,
-    kind: RealtimeAudioDeviceKind,
-    name: &str,
-) -> Option<cpal::Device> {
-    let devices = devices(host, kind).ok()?;
-    devices
-        .into_iter()
-        .find(|device| device.name().ok().as_deref() == Some(name))
-}
-
-fn devices(host: &cpal::Host, kind: RealtimeAudioDeviceKind) -> Result<Vec<cpal::Device>, String> {
+fn devices(host: &cpal::Host, kind: VoiceAudioDeviceKind) -> Result<Vec<cpal::Device>, String> {
     match kind {
-        RealtimeAudioDeviceKind::Microphone => host
+        VoiceAudioDeviceKind::Microphone => host
             .input_devices()
             .map(|devices| devices.collect())
             .map_err(|err| format!("failed to enumerate input audio devices: {err}")),
-        RealtimeAudioDeviceKind::Speaker => host
+        VoiceAudioDeviceKind::Speaker => host
             .output_devices()
             .map(|devices| devices.collect())
             .map_err(|err| format!("failed to enumerate output audio devices: {err}")),
     }
 }
 
-fn default_device(host: &cpal::Host, kind: RealtimeAudioDeviceKind) -> Option<cpal::Device> {
+fn default_device(host: &cpal::Host, kind: VoiceAudioDeviceKind) -> Option<cpal::Device> {
     match kind {
-        RealtimeAudioDeviceKind::Microphone => host.default_input_device(),
-        RealtimeAudioDeviceKind::Speaker => host.default_output_device(),
+        VoiceAudioDeviceKind::Microphone => host.default_input_device(),
+        VoiceAudioDeviceKind::Speaker => host.default_output_device(),
     }
 }
 
 fn default_config(
     device: &cpal::Device,
-    kind: RealtimeAudioDeviceKind,
+    kind: VoiceAudioDeviceKind,
 ) -> Result<cpal::SupportedStreamConfig, String> {
     match kind {
-        RealtimeAudioDeviceKind::Microphone => device
+        VoiceAudioDeviceKind::Microphone => device
             .default_input_config()
             .map_err(|err| format!("failed to get default input config: {err}")),
-        RealtimeAudioDeviceKind::Speaker => device
+        VoiceAudioDeviceKind::Speaker => device
             .default_output_config()
             .map_err(|err| format!("failed to get default output config: {err}")),
     }
 }
 
 fn preferred_input_sample_rate(range: &cpal::SupportedStreamConfigRange) -> cpal::SampleRate {
-    let min = range.min_sample_rate().0;
-    let max = range.max_sample_rate().0;
+    let min = range.min_sample_rate();
+    let max = range.max_sample_rate();
     if (min..=max).contains(&PREFERRED_INPUT_SAMPLE_RATE) {
-        cpal::SampleRate(PREFERRED_INPUT_SAMPLE_RATE)
+        PREFERRED_INPUT_SAMPLE_RATE
     } else if PREFERRED_INPUT_SAMPLE_RATE < min {
-        cpal::SampleRate(min)
+        min
     } else {
-        cpal::SampleRate(max)
+        max
     }
 }
 
-fn missing_device_error(kind: RealtimeAudioDeviceKind, configured_name: Option<&str>) -> String {
-    match (kind, configured_name) {
-        (RealtimeAudioDeviceKind::Microphone, Some(name)) => {
+fn missing_device_error(kind: VoiceAudioDeviceKind, configured_device_id: Option<&str>) -> String {
+    match (kind, configured_device_id) {
+        (VoiceAudioDeviceKind::Microphone, Some(device_id)) => {
             format!(
-                "configured microphone `{name}` was unavailable and no default input audio device was found"
+                "configured microphone `{device_id}` was unavailable and no default input audio device was found"
             )
         }
-        (RealtimeAudioDeviceKind::Speaker, Some(name)) => {
+        (VoiceAudioDeviceKind::Speaker, Some(device_id)) => {
             format!(
-                "configured speaker `{name}` was unavailable and no default output audio device was found"
+                "configured speaker `{device_id}` was unavailable and no default output audio device was found"
             )
         }
-        (RealtimeAudioDeviceKind::Microphone, None) => {
-            "no input audio device available".to_string()
-        }
-        (RealtimeAudioDeviceKind::Speaker, None) => "no output audio device available".to_string(),
+        (VoiceAudioDeviceKind::Microphone, None) => "no input audio device available".to_string(),
+        (VoiceAudioDeviceKind::Speaker, None) => "no output audio device available".to_string(),
     }
 }

codex-rs/tui/src/bottom_pane/chat_composer.rs

@@ -216,15 +216,12 @@ use std::collections::VecDeque;
 use std::ops::Range;
 use std::path::PathBuf;
 use std::sync::Arc;
-#[cfg(not(target_os = "linux"))]
 use std::sync::Mutex;
 use std::sync::atomic::AtomicBool;
 use std::sync::atomic::Ordering;
-#[cfg(not(target_os = "linux"))]
 use std::thread;
 use std::time::Duration;
 use std::time::Instant;
-#[cfg(not(target_os = "linux"))]
 use tokio::runtime::Handle;
 /// If the pasted content exceeds this number of characters, replace it with a
 /// placeholder in the UI.
@@ -321,13 +318,9 @@ struct VoiceState {
     space_hold_trigger: Option<Arc<AtomicBool>>,
     key_release_supported: bool,
     space_hold_repeat_seen: bool,
-    #[cfg(not(target_os = "linux"))]
     voice: Option<crate::voice::VoiceCapture>,
-    #[cfg(not(target_os = "linux"))]
     recording_placeholder_id: Option<String>,
-    #[cfg(not(target_os = "linux"))]
     space_recording_started_at: Option<Instant>,
-    #[cfg(not(target_os = "linux"))]
     space_recording_last_repeat_at: Option<Instant>,
 }
 
@@ -381,7 +374,6 @@ pub(crate) struct ChatComposer {
     footer_flash: Option<FooterFlash>,
     context_window_percent: Option<i64>,
     // Monotonically increasing identifier for textarea elements we insert.
-    #[cfg(not(target_os = "linux"))]
     next_element_id: u64,
     context_window_used_tokens: Option<i64>,
     skills: Option<Vec<SkillMetadata>>,
@@ -396,6 +388,7 @@ pub(crate) struct ChatComposer {
     personality_command_enabled: bool,
     realtime_conversation_enabled: bool,
     audio_device_selection_enabled: bool,
+    audio_microphone: Option<String>,
     windows_degraded_sandbox_active: bool,
     status_line_value: Option<Line<'static>>,
     status_line_enabled: bool,
@@ -487,7 +480,6 @@ impl ChatComposer {
             selected_remote_image_index: None,
             footer_flash: None,
             context_window_percent: None,
-            #[cfg(not(target_os = "linux"))]
             next_element_id: 0,
             context_window_used_tokens: None,
             skills: None,
@@ -502,6 +494,7 @@ impl ChatComposer {
             personality_command_enabled: false,
             realtime_conversation_enabled: false,
             audio_device_selection_enabled: false,
+            audio_microphone: None,
             windows_degraded_sandbox_active: false,
             status_line_value: None,
             status_line_enabled: false,
@@ -511,7 +504,6 @@ impl ChatComposer {
         this
     }
 
-    #[cfg(not(target_os = "linux"))]
     fn next_id(&mut self) -> String {
         let id = self.next_element_id;
         self.next_element_id = self.next_element_id.wrapping_add(1);
@@ -583,6 +575,10 @@ impl ChatComposer {
         self.audio_device_selection_enabled = enabled;
     }
 
+    pub fn set_audio_microphone(&mut self, microphone: Option<String>) {
+        self.audio_microphone = microphone;
+    }
+
     /// Compatibility shim for tests that still toggle the removed steer mode flag.
     #[cfg(test)]
     pub fn set_steer_enabled(&mut self, _enabled: bool) {}
@@ -598,9 +594,8 @@ impl ChatComposer {
         }
     }
 
-    #[cfg(not(target_os = "linux"))]
     fn voice_transcription_enabled(&self) -> bool {
-        self.voice_state.transcription_enabled && cfg!(not(target_os = "linux"))
+        self.voice_state.transcription_enabled
     }
     /// Centralized feature gating keeps config checks out of call sites.
     fn popups_enabled(&self) -> bool {
@@ -671,7 +666,6 @@ impl ChatComposer {
         }
 
         // Hide the cursor while recording voice input.
-        #[cfg(not(target_os = "linux"))]
         if self.voice_state.voice.is_some() {
             return None;
         }
@@ -732,7 +726,6 @@ impl ChatComposer {
     /// In all cases, clears any paste-burst Enter suppression state so a real paste cannot affect
     /// the next user Enter key, then syncs popup state.
     pub fn handle_paste(&mut self, pasted: String) -> bool {
-        #[cfg(not(target_os = "linux"))]
         if self.voice_state.voice.is_some() {
             return false;
         }
@@ -980,7 +973,6 @@ impl ChatComposer {
         local_image_paths: Vec<PathBuf>,
         mention_bindings: Vec<MentionBinding>,
     ) {
-        #[cfg(not(target_os = "linux"))]
         self.stop_all_transcription_spinners();
 
         // Clear any existing content, placeholders, and attachments first.
@@ -2760,15 +2752,6 @@ impl ChatComposer {
         }
     }
 
-    #[cfg(target_os = "linux")]
-    fn handle_voice_space_key_event(
-        &mut self,
-        _key_event: &KeyEvent,
-    ) -> Option<(InputResult, bool)> {
-        None
-    }
-
-    #[cfg(not(target_os = "linux"))]
     fn handle_voice_space_key_event(
         &mut self,
         key_event: &KeyEvent,
@@ -2843,15 +2826,6 @@ impl ChatComposer {
         }
     }
 
-    #[cfg(target_os = "linux")]
-    fn handle_key_event_while_recording(
-        &mut self,
-        _key_event: KeyEvent,
-    ) -> Option<(InputResult, bool)> {
-        None
-    }
-
-    #[cfg(not(target_os = "linux"))]
     fn handle_key_event_while_recording(
         &mut self,
         key_event: KeyEvent,
@@ -3624,7 +3598,6 @@ impl ChatComposer {
         self.has_focus = has_focus;
     }
 
-    #[cfg(not(target_os = "linux"))]
     pub(crate) fn is_recording(&self) -> bool {
         self.voice_state.voice.is_some()
     }
@@ -3662,7 +3635,6 @@ impl ChatComposer {
         }
     }
 
-    #[cfg(not(target_os = "linux"))]
     fn schedule_space_hold_timer(flag: Arc<AtomicBool>, frame: Option<FrameRequester>) {
         const HOLD_DELAY_MILLIS: u64 = 500;
         if let Ok(handle) = Handle::try_current() {
@@ -3680,7 +3652,6 @@ impl ChatComposer {
         }
     }
 
-    #[cfg(not(target_os = "linux"))]
     fn complete_space_hold_timer(flag: Arc<AtomicBool>, frame: Option<FrameRequester>) {
         flag.store(true, Ordering::Relaxed);
         if let Some(frame) = frame {
@@ -3705,7 +3676,6 @@ impl ChatComposer {
     }
 }
 
-#[cfg(not(target_os = "linux"))]
 impl ChatComposer {
     pub(crate) fn process_space_hold_trigger(&mut self) {
         if self.voice_transcription_enabled()
@@ -3849,7 +3819,7 @@ impl ChatComposer {
     /// Start voice capture and insert a placeholder element for the live meter.
     /// Returns true if recording began and UI should redraw; false on failure.
     fn start_recording_with_placeholder(&mut self) -> bool {
-        match crate::voice::VoiceCapture::start() {
+        match crate::voice::VoiceCapture::start(self.audio_microphone.as_deref()) {
             Ok(vc) => {
                 self.voice_state.voice = Some(vc);
                 if self.voice_state.key_release_supported {
@@ -3973,7 +3943,6 @@ impl ChatComposer {
         self.textarea.update_named_element_by_id(id, text)
     }
 
-    #[cfg(not(target_os = "linux"))]
     pub fn insert_transcription_placeholder(&mut self, text: &str) -> String {
         let id = self.next_id();
         self.textarea.insert_named_element(text, id.clone());
@@ -6461,7 +6430,6 @@ mod tests {
         assert!(!composer.voice_state.space_hold_repeat_seen);
     }
 
-    #[cfg(not(target_os = "linux"))]
     #[test]
     fn space_hold_timeout_without_release_or_repeat_keeps_typed_space() {
         let (tx, _rx) = unbounded_channel::<AppEvent>();
@@ -6493,7 +6461,6 @@ mod tests {
         assert!(!composer.voice_state.space_hold_repeat_seen);
     }
 
-    #[cfg(not(target_os = "linux"))]
     #[test]
     fn space_hold_timeout_with_repeat_uses_hold_path_without_release() {
         let (tx, _rx) = unbounded_channel::<AppEvent>();
@@ -6527,7 +6494,6 @@ mod tests {
         }
     }
 
-    #[cfg(not(target_os = "linux"))]
     #[test]
     fn space_hold_timeout_with_repeat_does_not_duplicate_existing_space() {
         let (tx, _rx) = unbounded_channel::<AppEvent>();
@@ -6561,7 +6527,6 @@ mod tests {
         }
     }
 
-    #[cfg(not(target_os = "linux"))]
     #[test]
     fn replace_transcription_stops_spinner_for_placeholder() {
         let (tx, _rx) = unbounded_channel::<AppEvent>();
@@ -6588,7 +6553,6 @@ mod tests {
         assert_eq!(composer.textarea.text(), "transcribed text");
     }
 
-    #[cfg(not(target_os = "linux"))]
     #[test]
     fn set_text_content_stops_all_transcription_spinners() {
         let (tx, _rx) = unbounded_channel::<AppEvent>();

codex-rs/tui/src/bottom_pane/mod.rs

@@ -303,6 +303,11 @@ impl BottomPane {
         self.request_redraw();
     }
 
+    pub fn set_audio_microphone(&mut self, microphone: Option<String>) {
+        self.composer.set_audio_microphone(microphone);
+        self.request_redraw();
+    }
+
     pub fn set_voice_transcription_enabled(&mut self, enabled: bool) {
         self.composer.set_voice_transcription_enabled(enabled);
         self.request_redraw();
@@ -346,7 +351,6 @@ impl BottomPane {
     pub fn handle_key_event(&mut self, key_event: KeyEvent) -> InputResult {
         // Do not globally intercept space; only composer handles hold-to-talk.
         // While recording, route all keys to the composer so it can stop on release or next key.
-        #[cfg(not(target_os = "linux"))]
         if self.composer.is_recording() {
             let (_ir, needs_redraw) = self.composer.handle_key_event(key_event);
             if needs_redraw {
@@ -480,7 +484,6 @@ impl BottomPane {
     // Space hold timeout is handled inside ChatComposer via an internal timer.
     pub(crate) fn pre_draw_tick(&mut self) {
         // Allow composer to process any time-based transitions before drawing
-        #[cfg(not(target_os = "linux"))]
         self.composer.process_space_hold_trigger();
         self.composer.sync_popups();
     }
@@ -1039,7 +1042,6 @@ impl BottomPane {
     }
 }
 
-#[cfg(not(target_os = "linux"))]
 impl BottomPane {
     pub(crate) fn insert_transcription_placeholder(&mut self, text: &str) -> String {
         let id = self.composer.insert_transcription_placeholder(text);

codex-rs/tui/src/bottom_pane/textarea.rs

@@ -894,7 +894,6 @@ impl TextArea {
         id
     }
 
-    #[cfg(not(target_os = "linux"))]
     pub fn insert_named_element(&mut self, text: &str, id: String) {
         let start = self.clamp_pos_for_insertion(self.cursor_pos);
         self.insert_str_at(start, text);

codex-rs/tui/src/chatwidget.rs

@@ -37,9 +37,9 @@ use std::sync::atomic::Ordering;
 use std::time::Duration;
 use std::time::Instant;
 
-use crate::app_event::RealtimeAudioDeviceKind;
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
-use crate::audio_device::list_realtime_audio_device_names;
+use crate::app_event::VoiceAudioDeviceKind;
+use crate::audio_device::VoiceAudioDevice;
+use crate::audio_device::list_voice_audio_devices;
 use crate::bottom_pane::StatusLineItem;
 use crate::bottom_pane::StatusLineSetupView;
 use crate::status::RateLimitWindowDisplay;
@@ -858,11 +858,10 @@ enum ReplayKind {
 impl ChatWidget {
     fn realtime_conversation_enabled(&self) -> bool {
         self.config.features.enabled(Feature::RealtimeConversation)
-            && cfg!(not(target_os = "linux"))
     }
 
     fn realtime_audio_device_selection_enabled(&self) -> bool {
-        self.realtime_conversation_enabled() && cfg!(feature = "voice-input")
+        self.realtime_conversation_enabled()
     }
 
     /// Synchronize the bottom-pane "task running" indicator with the current lifecycles.
@@ -2905,6 +2904,9 @@ impl ChatWidget {
         widget.bottom_pane.set_voice_transcription_enabled(
             widget.config.features.enabled(Feature::VoiceTranscription),
         );
+        widget
+            .bottom_pane
+            .set_audio_microphone(widget.config.realtime_audio.microphone.clone());
         widget
             .bottom_pane
             .set_realtime_conversation_enabled(widget.realtime_conversation_enabled());
@@ -3084,6 +3086,9 @@ impl ChatWidget {
         widget.bottom_pane.set_voice_transcription_enabled(
             widget.config.features.enabled(Feature::VoiceTranscription),
         );
+        widget
+            .bottom_pane
+            .set_audio_microphone(widget.config.realtime_audio.microphone.clone());
         widget
             .bottom_pane
             .set_realtime_conversation_enabled(widget.realtime_conversation_enabled());
@@ -3252,6 +3257,9 @@ impl ChatWidget {
         widget.bottom_pane.set_voice_transcription_enabled(
             widget.config.features.enabled(Feature::VoiceTranscription),
         );
+        widget
+            .bottom_pane
+            .set_audio_microphone(widget.config.realtime_audio.microphone.clone());
         widget
             .bottom_pane
             .set_realtime_conversation_enabled(widget.realtime_conversation_enabled());
@@ -5317,17 +5325,17 @@ impl ChatWidget {
 
     pub(crate) fn open_realtime_audio_popup(&mut self) {
         let items = [
-            RealtimeAudioDeviceKind::Microphone,
-            RealtimeAudioDeviceKind::Speaker,
+            VoiceAudioDeviceKind::Microphone,
+            VoiceAudioDeviceKind::Speaker,
         ]
         .into_iter()
         .map(|kind| {
             let description = Some(format!(
                 "Current: {}",
-                self.current_realtime_audio_selection_label(kind)
+                self.current_voice_audio_selection_label(kind)
             ));
             let actions: Vec<SelectionAction> = vec![Box::new(move |tx| {
-                tx.send(AppEvent::OpenRealtimeAudioDeviceSelection { kind });
+                tx.send(AppEvent::OpenVoiceAudioDeviceSelection { kind });
             })];
             SelectionItem {
                 name: kind.title().to_string(),
@@ -5348,52 +5356,43 @@ impl ChatWidget {
         });
     }
 
-    #[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
-    pub(crate) fn open_realtime_audio_device_selection(&mut self, kind: RealtimeAudioDeviceKind) {
-        match list_realtime_audio_device_names(kind) {
-            Ok(device_names) => {
-                self.open_realtime_audio_device_selection_with_names(kind, device_names);
+    pub(crate) fn open_voice_audio_device_selection(&mut self, kind: VoiceAudioDeviceKind) {
+        match list_voice_audio_devices(kind) {
+            Ok(devices) => {
+                self.open_voice_audio_device_selection_with_devices(kind, devices);
             }
             Err(err) => {
-                self.add_error_message(format!(
-                    "Failed to load realtime {} devices: {err}",
-                    kind.noun()
-                ));
+                self.add_error_message(format!("Failed to load {} devices: {err}", kind.noun()));
             }
         }
     }
 
-    #[cfg(any(target_os = "linux", not(feature = "voice-input")))]
-    pub(crate) fn open_realtime_audio_device_selection(&mut self, kind: RealtimeAudioDeviceKind) {
-        let _ = kind;
-    }
-
-    #[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
-    fn open_realtime_audio_device_selection_with_names(
+    fn open_voice_audio_device_selection_with_devices(
         &mut self,
-        kind: RealtimeAudioDeviceKind,
-        device_names: Vec<String>,
+        kind: VoiceAudioDeviceKind,
+        devices: Vec<VoiceAudioDevice>,
     ) {
-        let current_selection = self.current_realtime_audio_device_name(kind);
+        let current_selection = self.current_voice_audio_device_id(kind);
         let current_available = current_selection
             .as_deref()
-            .is_some_and(|name| device_names.iter().any(|device_name| device_name == name));
+            .is_some_and(|device_id| devices.iter().any(|device| device.id == device_id));
         let mut items = vec![SelectionItem {
             name: "System default".to_string(),
             description: Some("Use your operating system default device.".to_string()),
             is_current: current_selection.is_none(),
             actions: vec![Box::new(move |tx| {
-                tx.send(AppEvent::PersistRealtimeAudioDeviceSelection { kind, name: None });
+                tx.send(AppEvent::PersistVoiceAudioDeviceSelection {
+                    kind,
+                    device_id: None,
+                });
             })],
             dismiss_on_select: true,
             ..Default::default()
         }];
 
-        if let Some(selection) = current_selection.as_deref()
-            && !current_available
-        {
+        if current_selection.is_some() && !current_available {
             items.push(SelectionItem {
-                name: format!("Unavailable: {selection}"),
+                name: "Unavailable device".to_string(),
                 description: Some("Configured device is not currently available.".to_string()),
                 is_current: true,
                 is_disabled: true,
@@ -5402,17 +5401,32 @@ impl ChatWidget {
             });
         }
 
-        items.extend(device_names.into_iter().map(|device_name| {
-            let persisted_name = device_name.clone();
+        let duplicate_counts =
+            devices
+                .iter()
+                .fold(HashMap::<String, usize>::new(), |mut counts, device| {
+                    *counts.entry(device.label.clone()).or_default() += 1;
+                    counts
+                });
+
+        items.extend(devices.into_iter().map(|device| {
+            let persisted_device_id = device.id.clone();
+            let description = (duplicate_counts
+                .get(&device.label)
+                .copied()
+                .unwrap_or_default()
+                > 1)
+            .then(|| format!("Device ID: {}", Self::compact_device_id(&device.id)));
             let actions: Vec<SelectionAction> = vec![Box::new(move |tx| {
-                tx.send(AppEvent::PersistRealtimeAudioDeviceSelection {
+                tx.send(AppEvent::PersistVoiceAudioDeviceSelection {
                     kind,
-                    name: Some(persisted_name.clone()),
+                    device_id: Some(persisted_device_id.clone()),
                 });
             })];
             SelectionItem {
-                is_current: current_selection.as_deref() == Some(device_name.as_str()),
-                name: device_name,
+                is_current: current_selection.as_deref() == Some(device.id.as_str()),
+                name: device.label,
+                description,
                 actions,
                 dismiss_on_select: true,
                 ..Default::default()
@@ -5421,9 +5435,15 @@ impl ChatWidget {
 
         let mut header = ColumnRenderable::new();
         header.push(Line::from(format!("Select {}", kind.title()).bold()));
-        header.push(Line::from(
-            "Saved devices apply to realtime voice only.".dim(),
-        ));
+        let subtitle = match kind {
+            VoiceAudioDeviceKind::Microphone => {
+                "Saved microphone applies to transcription and realtime voice.".dim()
+            }
+            VoiceAudioDeviceKind::Speaker => {
+                "Saved speaker applies to realtime voice output.".dim()
+            }
+        };
+        header.push(Line::from(subtitle));
 
         self.bottom_pane.show_selection_view(SelectionViewParams {
             header: Box::new(header),
@@ -5433,9 +5453,9 @@ impl ChatWidget {
         });
     }
 
-    pub(crate) fn open_realtime_audio_restart_prompt(&mut self, kind: RealtimeAudioDeviceKind) {
+    pub(crate) fn open_voice_audio_restart_prompt(&mut self, kind: VoiceAudioDeviceKind) {
         let restart_actions: Vec<SelectionAction> = vec![Box::new(move |tx| {
-            tx.send(AppEvent::RestartRealtimeAudioDevice { kind });
+            tx.send(AppEvent::RestartVoiceAudioDevice { kind });
         })];
         let items = vec![
             SelectionItem {
@@ -6814,14 +6834,17 @@ impl ChatWidget {
         self.config.personality = Some(personality);
     }
 
-    pub(crate) fn set_realtime_audio_device(
+    pub(crate) fn set_voice_audio_device(
         &mut self,
-        kind: RealtimeAudioDeviceKind,
-        name: Option<String>,
+        kind: VoiceAudioDeviceKind,
+        device_id: Option<String>,
     ) {
         match kind {
-            RealtimeAudioDeviceKind::Microphone => self.config.realtime_audio.microphone = name,
-            RealtimeAudioDeviceKind::Speaker => self.config.realtime_audio.speaker = name,
+            VoiceAudioDeviceKind::Microphone => {
+                self.config.realtime_audio.microphone = device_id.clone();
+                self.bottom_pane.set_audio_microphone(device_id);
+            }
+            VoiceAudioDeviceKind::Speaker => self.config.realtime_audio.speaker = device_id,
         }
     }
 
@@ -6857,16 +6880,38 @@ impl ChatWidget {
         self.realtime_conversation.is_active()
     }
 
-    fn current_realtime_audio_device_name(&self, kind: RealtimeAudioDeviceKind) -> Option<String> {
+    fn current_voice_audio_device_id(&self, kind: VoiceAudioDeviceKind) -> Option<String> {
         match kind {
-            RealtimeAudioDeviceKind::Microphone => self.config.realtime_audio.microphone.clone(),
-            RealtimeAudioDeviceKind::Speaker => self.config.realtime_audio.speaker.clone(),
+            VoiceAudioDeviceKind::Microphone => self.config.realtime_audio.microphone.clone(),
+            VoiceAudioDeviceKind::Speaker => self.config.realtime_audio.speaker.clone(),
         }
     }
 
-    fn current_realtime_audio_selection_label(&self, kind: RealtimeAudioDeviceKind) -> String {
-        self.current_realtime_audio_device_name(kind)
-            .unwrap_or_else(|| "System default".to_string())
+    fn current_voice_audio_selection_label(&self, kind: VoiceAudioDeviceKind) -> String {
+        let Some(device_id) = self.current_voice_audio_device_id(kind) else {
+            return "System default".to_string();
+        };
+
+        list_voice_audio_devices(kind)
+            .ok()
+            .and_then(|devices| {
+                devices
+                    .into_iter()
+                    .find(|device| device.id == device_id)
+                    .map(|device| device.label)
+            })
+            .unwrap_or_else(|| "Unavailable device".to_string())
+    }
+
+    fn compact_device_id(device_id: &str) -> String {
+        const MAX_LEN: usize = 18;
+        if device_id.len() <= MAX_LEN {
+            return device_id.to_string();
+        }
+
+        let start = &device_id[..8];
+        let end = &device_id[device_id.len() - 8..];
+        format!("{start}...{end}")
     }
 
     fn sync_personality_command_enabled(&mut self) {
@@ -7922,7 +7967,6 @@ impl ChatWidget {
     }
 }
 
-#[cfg(not(target_os = "linux"))]
 impl ChatWidget {
     pub(crate) fn replace_transcription(&mut self, id: &str, text: &str) {
         self.bottom_pane.replace_transcription(id, text);

codex-rs/tui/src/chatwidget/realtime.rs

@@ -24,11 +24,8 @@ pub(super) struct RealtimeConversationUiState {
     session_id: Option<String>,
     warned_audio_only_submission: bool,
     meter_placeholder_id: Option<String>,
-    #[cfg(not(target_os = "linux"))]
     capture_stop_flag: Option<Arc<AtomicBool>>,
-    #[cfg(not(target_os = "linux"))]
     capture: Option<crate::voice::VoiceCapture>,
-    #[cfg(not(target_os = "linux"))]
     audio_player: Option<crate::voice::RealtimeAudioPlayer>,
 }
 
@@ -207,25 +204,17 @@ impl ChatWidget {
     }
 
     fn enqueue_realtime_audio_out(&mut self, frame: &RealtimeAudioFrame) {
-        #[cfg(not(target_os = "linux"))]
-        {
-            if self.realtime_conversation.audio_player.is_none() {
-                self.realtime_conversation.audio_player =
-                    crate::voice::RealtimeAudioPlayer::start(&self.config).ok();
-            }
-            if let Some(player) = &self.realtime_conversation.audio_player
-                && let Err(err) = player.enqueue_frame(frame)
-            {
-                warn!("failed to play realtime audio: {err}");
-            }
+        if self.realtime_conversation.audio_player.is_none() {
+            self.realtime_conversation.audio_player =
+                crate::voice::RealtimeAudioPlayer::start(&self.config).ok();
         }
-        #[cfg(target_os = "linux")]
+        if let Some(player) = &self.realtime_conversation.audio_player
+            && let Err(err) = player.enqueue_frame(frame)
         {
-            let _ = frame;
+            warn!("failed to play realtime audio: {err}");
         }
     }
 
-    #[cfg(not(target_os = "linux"))]
     fn start_realtime_local_audio(&mut self) {
         if self.realtime_conversation.capture_stop_flag.is_some() {
             return;
@@ -279,21 +268,17 @@ impl ChatWidget {
         });
     }
 
-    #[cfg(target_os = "linux")]
-    fn start_realtime_local_audio(&mut self) {}
-
-    #[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
-    pub(crate) fn restart_realtime_audio_device(&mut self, kind: RealtimeAudioDeviceKind) {
+    pub(crate) fn restart_voice_audio_device(&mut self, kind: VoiceAudioDeviceKind) {
         if !self.realtime_conversation.is_active() {
             return;
         }
 
         match kind {
-            RealtimeAudioDeviceKind::Microphone => {
+            VoiceAudioDeviceKind::Microphone => {
                 self.stop_realtime_microphone();
                 self.start_realtime_local_audio();
             }
-            RealtimeAudioDeviceKind::Speaker => {
+            VoiceAudioDeviceKind::Speaker => {
                 self.stop_realtime_speaker();
                 match crate::voice::RealtimeAudioPlayer::start(&self.config) {
                     Ok(player) => {
@@ -308,23 +293,11 @@ impl ChatWidget {
         self.request_redraw();
     }
 
-    #[cfg(any(target_os = "linux", not(feature = "voice-input")))]
-    pub(crate) fn restart_realtime_audio_device(&mut self, kind: RealtimeAudioDeviceKind) {
-        let _ = kind;
-    }
-
-    #[cfg(not(target_os = "linux"))]
     fn stop_realtime_local_audio(&mut self) {
         self.stop_realtime_microphone();
         self.stop_realtime_speaker();
     }
 
-    #[cfg(target_os = "linux")]
-    fn stop_realtime_local_audio(&mut self) {
-        self.realtime_conversation.meter_placeholder_id = None;
-    }
-
-    #[cfg(not(target_os = "linux"))]
     fn stop_realtime_microphone(&mut self) {
         if let Some(flag) = self.realtime_conversation.capture_stop_flag.take() {
             flag.store(true, Ordering::Relaxed);
@@ -337,7 +310,6 @@ impl ChatWidget {
         }
     }
 
-    #[cfg(not(target_os = "linux"))]
     fn stop_realtime_speaker(&mut self) {
         if let Some(player) = self.realtime_conversation.audio_player.take() {
             player.clear();

codex-rs/tui/src/chatwidget/snapshots/codex_tui__chatwidget__tests__realtime_microphone_picker_popup.snap

@@ -3,15 +3,14 @@ source: tui/src/chatwidget/tests.rs
 expression: popup
 ---
   Select Microphone
-  Saved devices apply to realtime voice only.
+  Saved microphone applies to transcription and realtime voice.
 
-  1. System default                                Use your operating system
-                                                   default device.
-› 2. Unavailable: Studio Mic (current) (disabled)  Configured device is not
-                                                   currently available.
-                                                   (disabled: Reconnect the
-                                                   device or choose another
-                                                   one.)
+  1. System default                           Use your operating system
+                                              default device.
+› 2. Unavailable device (current) (disabled)  Configured device is not
+                                              currently available. (disabled:
+                                              Reconnect the device or choose
+                                              another one.)
   3. Built-in Mic
   4. USB Mic
 

codex-rs/tui/src/chatwidget/tests.rs

@@ -7,9 +7,9 @@
 use super::*;
 use crate::app_event::AppEvent;
 use crate::app_event::ExitMode;
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
-use crate::app_event::RealtimeAudioDeviceKind;
+use crate::app_event::VoiceAudioDeviceKind;
 use crate::app_event_sender::AppEventSender;
+use crate::audio_device::VoiceAudioDevice;
 use crate::bottom_pane::FeedbackAudience;
 use crate::bottom_pane::LocalImageAttachment;
 use crate::bottom_pane::MentionBinding;
@@ -6011,7 +6011,6 @@ async fn personality_selection_popup_snapshot() {
     assert_snapshot!("personality_selection_popup", popup);
 }
 
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
 #[tokio::test]
 async fn realtime_audio_selection_popup_snapshot() {
     let (mut chat, _rx, _op_rx) = make_chatwidget_manual(Some("gpt-5.2-codex")).await;
@@ -6021,7 +6020,6 @@ async fn realtime_audio_selection_popup_snapshot() {
     assert_snapshot!("realtime_audio_selection_popup", popup);
 }
 
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
 #[tokio::test]
 async fn realtime_audio_selection_popup_narrow_snapshot() {
     let (mut chat, _rx, _op_rx) = make_chatwidget_manual(Some("gpt-5.2-codex")).await;
@@ -6031,27 +6029,43 @@ async fn realtime_audio_selection_popup_narrow_snapshot() {
     assert_snapshot!("realtime_audio_selection_popup_narrow", popup);
 }
 
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
 #[tokio::test]
 async fn realtime_microphone_picker_popup_snapshot() {
     let (mut chat, _rx, _op_rx) = make_chatwidget_manual(Some("gpt-5.2-codex")).await;
-    chat.config.realtime_audio.microphone = Some("Studio Mic".to_string());
-    chat.open_realtime_audio_device_selection_with_names(
-        RealtimeAudioDeviceKind::Microphone,
-        vec!["Built-in Mic".to_string(), "USB Mic".to_string()],
+    chat.config.realtime_audio.microphone = Some("mic-unavailable".to_string());
+    chat.open_voice_audio_device_selection_with_devices(
+        VoiceAudioDeviceKind::Microphone,
+        vec![
+            VoiceAudioDevice {
+                id: "mic-1".to_string(),
+                label: "Built-in Mic".to_string(),
+            },
+            VoiceAudioDevice {
+                id: "mic-2".to_string(),
+                label: "USB Mic".to_string(),
+            },
+        ],
     );
 
     let popup = render_bottom_popup(&chat, 80);
     assert_snapshot!("realtime_microphone_picker_popup", popup);
 }
 
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
 #[tokio::test]
 async fn realtime_audio_picker_emits_persist_event() {
     let (mut chat, mut rx, _op_rx) = make_chatwidget_manual(Some("gpt-5.2-codex")).await;
-    chat.open_realtime_audio_device_selection_with_names(
-        RealtimeAudioDeviceKind::Speaker,
-        vec!["Desk Speakers".to_string(), "Headphones".to_string()],
+    chat.open_voice_audio_device_selection_with_devices(
+        VoiceAudioDeviceKind::Speaker,
+        vec![
+            VoiceAudioDevice {
+                id: "speaker-1".to_string(),
+                label: "Desk Speakers".to_string(),
+            },
+            VoiceAudioDevice {
+                id: "speaker-2".to_string(),
+                label: "Headphones".to_string(),
+            },
+        ],
     );
 
     chat.handle_key_event(KeyEvent::new(KeyCode::Down, KeyModifiers::NONE));
@@ -6060,10 +6074,10 @@ async fn realtime_audio_picker_emits_persist_event() {
 
     assert_matches!(
         rx.try_recv(),
-        Ok(AppEvent::PersistRealtimeAudioDeviceSelection {
-            kind: RealtimeAudioDeviceKind::Speaker,
-            name: Some(name),
-        }) if name == "Headphones"
+        Ok(AppEvent::PersistVoiceAudioDeviceSelection {
+            kind: VoiceAudioDeviceKind::Speaker,
+            device_id: Some(device_id),
+        }) if device_id == "speaker-2"
     );
 }
 

codex-rs/tui/src/lib.rs

@@ -63,7 +63,6 @@ mod app_backtrack;
 mod app_event;
 mod app_event_sender;
 mod ascii_animation;
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
 mod audio_device;
 mod bottom_pane;
 mod chatwidget;
@@ -119,99 +118,7 @@ pub mod update_action;
 mod update_prompt;
 mod updates;
 mod version;
-#[cfg(all(not(target_os = "linux"), feature = "voice-input"))]
 mod voice;
-#[cfg(all(not(target_os = "linux"), not(feature = "voice-input")))]
-mod voice {
-    use crate::app_event::AppEvent;
-    use crate::app_event_sender::AppEventSender;
-    use codex_core::config::Config;
-    use codex_protocol::protocol::RealtimeAudioFrame;
-    use std::sync::Arc;
-    use std::sync::Mutex;
-    use std::sync::atomic::AtomicBool;
-    use std::sync::atomic::AtomicU16;
-
-    pub struct RecordedAudio {
-        pub data: Vec<i16>,
-        pub sample_rate: u32,
-        pub channels: u16,
-    }
-
-    pub struct VoiceCapture;
-
-    pub(crate) struct RecordingMeterState;
-
-    pub(crate) struct RealtimeAudioPlayer;
-
-    impl VoiceCapture {
-        pub fn start() -> Result<Self, String> {
-            Err("voice input is unavailable in this build".to_string())
-        }
-
-        pub fn start_realtime(_config: &Config, _tx: AppEventSender) -> Result<Self, String> {
-            Err("voice input is unavailable in this build".to_string())
-        }
-
-        pub fn stop(self) -> Result<RecordedAudio, String> {
-            Err("voice input is unavailable in this build".to_string())
-        }
-
-        pub fn data_arc(&self) -> Arc<Mutex<Vec<i16>>> {
-            Arc::new(Mutex::new(Vec::new()))
-        }
-
-        pub fn stopped_flag(&self) -> Arc<AtomicBool> {
-            Arc::new(AtomicBool::new(true))
-        }
-
-        pub fn sample_rate(&self) -> u32 {
-            0
-        }
-
-        pub fn channels(&self) -> u16 {
-            0
-        }
-
-        pub fn last_peak_arc(&self) -> Arc<AtomicU16> {
-            Arc::new(AtomicU16::new(0))
-        }
-    }
-
-    impl RecordingMeterState {
-        pub(crate) fn new() -> Self {
-            Self
-        }
-
-        pub(crate) fn next_text(&mut self, _peak: u16) -> String {
-            "⠤⠤⠤⠤".to_string()
-        }
-    }
-
-    impl RealtimeAudioPlayer {
-        pub(crate) fn start(_config: &Config) -> Result<Self, String> {
-            Err("voice output is unavailable in this build".to_string())
-        }
-
-        pub(crate) fn enqueue_frame(&self, _frame: &RealtimeAudioFrame) -> Result<(), String> {
-            Err("voice output is unavailable in this build".to_string())
-        }
-
-        pub(crate) fn clear(&self) {}
-    }
-
-    pub fn transcribe_async(
-        id: String,
-        _audio: RecordedAudio,
-        _context: Option<String>,
-        tx: AppEventSender,
-    ) {
-        tx.send(AppEvent::TranscriptionFailed {
-            id,
-            error: "voice input is unavailable in this build".to_string(),
-        });
-    }
-}
 mod wrapping;
 
 #[cfg(test)]

codex-rs/tui/src/voice.rs

@@ -11,7 +11,6 @@ use codex_protocol::protocol::ConversationAudioParams;
 use codex_protocol::protocol::Op;
 use codex_protocol::protocol::RealtimeAudioFrame;
 use cpal::traits::DeviceTrait;
-use cpal::traits::HostTrait;
 use cpal::traits::StreamTrait;
 use hound::SampleFormat;
 use hound::WavSpec;
@@ -54,10 +53,10 @@ pub struct VoiceCapture {
 }
 
 impl VoiceCapture {
-    pub fn start() -> Result<Self, String> {
-        let (device, config) = select_default_input_device_and_config()?;
+    pub fn start(device_id: Option<&str>) -> Result<Self, String> {
+        let (device, config) = select_input_device_and_config(device_id)?;
 
-        let sample_rate = config.sample_rate().0;
+        let sample_rate = config.sample_rate();
         let channels = config.channels();
         let data: Arc<Mutex<Vec<i16>>> = Arc::new(Mutex::new(Vec::new()));
         let stopped = Arc::new(AtomicBool::new(false));
@@ -81,7 +80,7 @@ impl VoiceCapture {
     pub fn start_realtime(config: &Config, tx: AppEventSender) -> Result<Self, String> {
         let (device, config) = select_realtime_input_device_and_config(config)?;
 
-        let sample_rate = config.sample_rate().0;
+        let sample_rate = config.sample_rate();
         let channels = config.channels();
         let data: Arc<Mutex<Vec<i16>>> = Arc::new(Mutex::new(Vec::new()));
         let stopped = Arc::new(AtomicBool::new(false));
@@ -266,14 +265,10 @@ pub fn transcribe_async(
 // Voice input helpers
 // -------------------------
 
-fn select_default_input_device_and_config()
--> Result<(cpal::Device, cpal::SupportedStreamConfig), String> {
-    let host = cpal::default_host();
-    let device = host
-        .default_input_device()
-        .ok_or_else(|| "no input audio device available".to_string())?;
-    let config = crate::audio_device::preferred_input_config(&device)?;
-    Ok((device, config))
+fn select_input_device_and_config(
+    device_id: Option<&str>,
+) -> Result<(cpal::Device, cpal::SupportedStreamConfig), String> {
+    crate::audio_device::select_input_device_and_config(device_id)
 }
 
 fn select_realtime_input_device_and_config(
@@ -493,7 +488,7 @@ impl RealtimeAudioPlayer {
     pub(crate) fn start(config: &Config) -> Result<Self, String> {
         let (device, config) =
             crate::audio_device::select_configured_output_device_and_config(config)?;
-        let output_sample_rate = config.sample_rate().0;
+        let output_sample_rate = config.sample_rate();
         let output_channels = config.channels();
         let queue = Arc::new(Mutex::new(VecDeque::new()));
         let stream = build_output_stream(&device, &config, Arc::clone(&queue))?;

docs/install.md

@@ -49,6 +49,15 @@ just test
 cargo test --all-features
 ```
 
+On Linux, building from source currently requires ALSA development headers because local voice support is built with `cpal`. On Debian/Ubuntu, install them with:
+
+```bash
+sudo apt-get update
+sudo apt-get install -y libasound2-dev
+```
+
+If you just want to run Codex on Linux, prefer the prebuilt GitHub Release binaries instead of building from source.
+
 ## Tracing / verbose logging
 
 Codex is written in Rust, so it honors the `RUST_LOG` environment variable to configure its logging behavior.