Add realtime voice selection (#17176)

- Add realtime voice selection for realtime/start.
- Expose the supported v1/v2 voice lists and cover explicit, configured,
default, and invalid voice paths.

codex-rs/app-server-protocol/src/protocol/common.rs

@@ -414,6 +414,11 @@ client_request_definitions! {
         params: v2::ThreadRealtimeStopParams,
         response: v2::ThreadRealtimeStopResponse,
     },
+    #[experimental("thread/realtime/listVoices")]
+    ThreadRealtimeListVoices => "thread/realtime/listVoices" {
+        params: v2::ThreadRealtimeListVoicesParams,
+        response: v2::ThreadRealtimeListVoicesResponse,
+    },
     ReviewStart => "review/start" {
         params: v2::ReviewStartParams,
         response: v2::ReviewStartResponse,
@@ -1764,6 +1769,7 @@ mod tests {
                 prompt: Some(Some("You are on a call".to_string())),
                 session_id: Some("sess_456".to_string()),
                 transport: None,
+                voice: Some(codex_protocol::protocol::RealtimeVoice::Marin),
             },
         };
         assert_eq!(
@@ -1774,7 +1780,8 @@ mod tests {
                     "threadId": "thr_123",
                     "prompt": "You are on a call",
                     "sessionId": "sess_456",
-                    "transport": null
+                    "transport": null,
+                    "voice": "marin"
                 }
             }),
             serde_json::to_value(&request)?,
@@ -1791,6 +1798,7 @@ mod tests {
                 prompt: None,
                 session_id: None,
                 transport: None,
+                voice: None,
             },
         };
         assert_eq!(
@@ -1800,7 +1808,8 @@ mod tests {
                 "params": {
                     "threadId": "thr_123",
                     "sessionId": null,
-                    "transport": null
+                    "transport": null,
+                    "voice": null
                 }
             }),
             serde_json::to_value(&default_prompt_request)?,
@@ -1813,6 +1822,7 @@ mod tests {
                 prompt: Some(None),
                 session_id: None,
                 transport: None,
+                voice: None,
             },
         };
         assert_eq!(
@@ -1823,7 +1833,8 @@ mod tests {
                     "threadId": "thr_123",
                     "prompt": null,
                     "sessionId": null,
-                    "transport": null
+                    "transport": null,
+                    "voice": null
                 }
             }),
             serde_json::to_value(&null_prompt_request)?,
@@ -1835,7 +1846,8 @@ mod tests {
             "params": {
                 "threadId": "thr_123",
                 "sessionId": null,
-                "transport": null
+                "transport": null,
+                "voice": null
             }
         });
         assert_eq!(
@@ -1850,7 +1862,8 @@ mod tests {
                 "threadId": "thr_123",
                 "prompt": null,
                 "sessionId": null,
-                "transport": null
+                "transport": null,
+                "voice": null
             }
         });
         assert_eq!(
@@ -1934,6 +1947,7 @@ mod tests {
                 prompt: Some(Some("You are on a call".to_string())),
                 session_id: None,
                 transport: None,
+                voice: None,
             },
         };
         let reason = crate::experimental_api::ExperimentalApi::experimental_reason(&request);

codex-rs/app-server-protocol/src/protocol/v2.rs

@@ -72,6 +72,8 @@ use codex_protocol::protocol::RateLimitWindow as CoreRateLimitWindow;
 use codex_protocol::protocol::ReadOnlyAccess as CoreReadOnlyAccess;
 use codex_protocol::protocol::RealtimeAudioFrame as CoreRealtimeAudioFrame;
 use codex_protocol::protocol::RealtimeConversationVersion;
+use codex_protocol::protocol::RealtimeVoice;
+use codex_protocol::protocol::RealtimeVoicesList;
 use codex_protocol::protocol::ReviewDecision as CoreReviewDecision;
 use codex_protocol::protocol::SessionSource as CoreSessionSource;
 use codex_protocol::protocol::SkillDependencies as CoreSkillDependencies;
@@ -3866,6 +3868,8 @@ pub struct ThreadRealtimeStartParams {
     pub session_id: Option<String>,
     #[ts(optional = nullable)]
     pub transport: Option<ThreadRealtimeStartTransport>,
+    #[ts(optional = nullable)]
+    pub voice: Option<RealtimeVoice>,
 }
 
 /// EXPERIMENTAL - transport used by thread realtime.
@@ -3931,6 +3935,20 @@ pub struct ThreadRealtimeStopParams {
 #[ts(export_to = "v2/")]
 pub struct ThreadRealtimeStopResponse {}
 
+/// EXPERIMENTAL - list voices supported by thread realtime.
+#[derive(Serialize, Deserialize, Debug, Default, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ThreadRealtimeListVoicesParams {}
+
+/// EXPERIMENTAL - response for listing supported realtime voices.
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ThreadRealtimeListVoicesResponse {
+    pub voices: RealtimeVoicesList,
+}
+
 /// EXPERIMENTAL - emitted when thread realtime startup is accepted.
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]