Let voice mode TTS use the model and voice an OpenAI-compatible server serves (#6060)

* fix(server): accept voice and model names from self-hosted speech servers

features.voiceMode.tts.voice and TTS_VOICE were validated against the
OpenAI voice catalog, so a server whose voices live under other names
could not be configured at all: config.json failed to parse and an env
var threw inside the speech config resolver, which config.ts calls
without a catch, so the daemon refused to start. TTS_MODEL had the same
problem for model ids the server serves itself. Both are free-form now,
with model ids keeping their case because rosters use values like
kokoro-v1:Q8_0. Regenerate the config JSON schema.

* fix(server): send TTS voice names as written and keep the config schema current

A self-hosted OpenAI-compatible server owns its voice ids and their case, so
lowercasing the configured voice could turn a valid id into one the server
does not know. Keep the published config schema in its existing form and
change only the voice field.

Co-authored-by: khimaros <me@khimaros.com>

* fix(server): drop the TTS baseUrl from the init log

The configured baseUrl can carry a token in its query string, and logging
it is not needed to accept self-hosted voice and model names.

* test(server): await bounded worktree fixture cleanup on Windows

---------

Co-authored-by: khimaros <me@khimaros.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>

packages/server/.env.example

@@ -11,9 +11,10 @@ OPENAI_TTS_BASE_URL=
 
 # TTS Configuration (optional - defaults shown)
 TTS_VOICE=alloy
-# Available voices: alloy, echo, fable, onyx, nova, shimmer
+# Any voice the server knows: OpenAI ships alloy, echo, fable, onyx, nova, shimmer
 TTS_MODEL=tts-1
-# Available models: tts-1 (faster), tts-1-hd (higher quality)
+# OpenAI ships tts-1 (faster) and tts-1-hd (higher quality); self-hosted
+# servers accept whatever model id they serve. Case sensitive.
 
 # Server Configuration
 # Location for runtime state (agents.json, etc.). Defaults to ~/.paseo

packages/server/src/server/persisted-config.test.ts

@@ -674,6 +674,19 @@ describe("PersistedConfigSchema voice mode config", () => {
     expect(parsed.features?.dictation?.stt?.language).toBe("fr");
     expect(parsed.features?.voiceMode?.stt?.language).toBe("de");
   });
+
+  test("accepts a voice and model from a self-hosted OpenAI-compatible server", () => {
+    const parsed = PersistedConfigSchema.parse({
+      features: {
+        voiceMode: {
+          tts: { provider: "openai", model: "kokoro-v1:Q8_0", voice: "aiden" },
+        },
+      },
+    });
+
+    expect(parsed.features?.voiceMode?.tts?.model).toBe("kokoro-v1:Q8_0");
+    expect(parsed.features?.voiceMode?.tts?.voice).toBe("aiden");
+  });
 });
 
 describe("loadPersistedConfig", () => {

packages/server/src/server/persisted-config.ts

@@ -144,7 +144,8 @@ const FeatureVoiceModeSchema = z
       .object({
         provider: SpeechProviderIdSchema.optional(),
         model: z.string().min(1).optional(),
-        voice: z.enum(["alloy", "echo", "fable", "onyx", "nova", "shimmer"]).optional(),
+        // Free-form: self-hosted OpenAI-compatible servers expose their own voice names.
+        voice: z.string().trim().min(1).optional(),
         speakerId: z.number().int().optional(),
         speed: z.number().optional(),
       })

packages/server/src/server/speech/providers/openai/config.test.ts

@@ -203,4 +203,40 @@ describe("resolveOpenAiSpeechConfig", () => {
     expect(resolved?.stt?.apiKey).toBe("stt-only-key");
     expect(resolved?.tts).toBeUndefined();
   });
+
+  test("sends the model and voice an OpenAI-compatible server serves, as written", () => {
+    const persisted = PersistedConfigSchema.parse({
+      providers: {
+        openai: { tts: { apiKey: "tts-key", baseUrl: "https://tts.example.com/v1" } },
+      },
+      features: {
+        voiceMode: { tts: { provider: "openai", model: "kokoro-v1:Q8_0", voice: "Aiden" } },
+      },
+    });
+
+    const resolved = resolveOpenAiSpeechConfig({
+      env: {} as NodeJS.ProcessEnv,
+      persisted,
+      providers: ALL_OPENAI,
+    });
+
+    expect(resolved?.tts?.model).toBe("kokoro-v1:Q8_0");
+    expect(resolved?.tts?.voice).toBe("Aiden");
+  });
+
+  test("accepts OpenAI's own gpt-4o-mini-tts model", () => {
+    const persisted = PersistedConfigSchema.parse({
+      providers: { openai: { tts: { apiKey: "tts-key" } } },
+      features: { voiceMode: { tts: { provider: "openai", model: "gpt-4o-mini-tts" } } },
+    });
+
+    const resolved = resolveOpenAiSpeechConfig({
+      env: {} as NodeJS.ProcessEnv,
+      persisted,
+      providers: ALL_OPENAI,
+    });
+
+    expect(resolved?.tts?.model).toBe("gpt-4o-mini-tts");
+    expect(resolved?.tts?.voice).toBe("alloy");
+  });
 });

packages/server/src/server/speech/providers/openai/config.ts

@@ -12,10 +12,6 @@ export interface OpenAiSpeechProviderConfig {
   tts?: Partial<TTSConfig> & { apiKey?: string };
 }
 
-const OpenAiTtsVoiceSchema = z.enum(["alloy", "echo", "fable", "onyx", "nova", "shimmer"]);
-
-const OpenAiTtsModelSchema = z.enum(["tts-1", "tts-1-hd"]);
-
 const NumberLikeSchema = z.union([z.number(), z.string().trim().min(1)]);
 
 const OptionalFiniteNumberSchema = NumberLikeSchema.pipe(
@@ -44,14 +40,12 @@ const OpenAiSttOptionsSchema = z.object({
   sttModel: OptionalTrimmedStringSchema,
 });
 
+// Voice and model are free-form so a self-hosted server can use its own ids; the
+// OpenAI catalog is only the default. Both are sent as written, because the
+// server owns those ids and their case (rosters use ids like kokoro-v1:Q8_0).
 const OpenAiTtsOptionsSchema = z.object({
-  ttsVoice: z.string().trim().toLowerCase().pipe(OpenAiTtsVoiceSchema).default("alloy"),
-  ttsModel: z
-    .string()
-    .trim()
-    .toLowerCase()
-    .pipe(OpenAiTtsModelSchema)
-    .default(DEFAULT_OPENAI_TTS_MODEL),
+  ttsVoice: z.string().trim().default("alloy"),
+  ttsModel: z.string().trim().default(DEFAULT_OPENAI_TTS_MODEL),
 });
 
 function isOpenAiProviderActive(provider: { enabled?: boolean; provider: string }): boolean {

packages/server/src/server/speech/providers/openai/tts.ts

@@ -8,8 +8,8 @@ export type { SpeechStreamResult };
 export interface TTSConfig {
   apiKey: string;
   baseUrl?: string;
-  model?: "tts-1" | "tts-1-hd";
-  voice?: "alloy" | "echo" | "fable" | "onyx" | "nova" | "shimmer";
+  model?: "tts-1" | "tts-1-hd" | (string & {});
+  voice?: "alloy" | "echo" | "fable" | "onyx" | "nova" | "shimmer" | (string & {});
   responseFormat?: "mp3" | "opus" | "aac" | "flac" | "wav" | "pcm";
 }
 

packages/server/src/utils/worktree.test.ts

@@ -10,6 +10,7 @@ import {
   type WorktreeConfig,
 } from "./worktree";
 import { execFileSync } from "child_process";
+import { rm } from "node:fs/promises";
 import {
   mkdtempSync,
   mkdirSync,
@@ -89,8 +90,8 @@ describe("paseo worktree manager", () => {
     });
   });
 
-  afterEach(() => {
-    rmSync(tempDir, { recursive: true, force: true });
+  afterEach(async () => {
+    await rm(tempDir, { recursive: true, force: true, maxRetries: 10, retryDelay: 100 });
   });
 
   it("treats a worktree as paseo-owned even when its .git admin is missing", async () => {

packages/website/public/schemas/paseo.config.v1.json

@@ -711,7 +711,7 @@
                 },
                 "voice": {
                   "type": "string",
-                  "enum": ["alloy", "echo", "fable", "onyx", "nova", "shimmer"]
+                  "minLength": 1
                 },
                 "speakerId": {
                   "type": "integer",