Add custom voice input fallback for non-standard TTS providers like DeepInfra

src/app/api/tts/voices/route.ts

@@ -21,10 +21,10 @@ export async function GET(req: NextRequest) {
     }
 
     const data = await response.json();
-    return NextResponse.json({ voices: data.voices || DEFAULT_VOICES });
+    return NextResponse.json({ voices: data.voices || DEFAULT_VOICES, failed: false });
   } catch (error) {
     console.error('Error fetching voices:', error);
-    // Return default voices on error
+    // Return default voices on error with failed flag
-    return NextResponse.json({ voices: DEFAULT_VOICES });
+    return NextResponse.json({ voices: DEFAULT_VOICES, failed: true });
   }
 }

src/components/player/TTSPlayer.tsx

@@ -27,6 +27,7 @@ export default function TTSPlayer({ currentPage, numPages }: {
     setAudioPlayerSpeedAndRestart,
     setVoiceAndRestart,
     availableVoices,
+    voiceApiFailed,
     skipToLocation,
   } = useTTS();
 
@@ -77,7 +78,7 @@ export default function TTSPlayer({ currentPage, numPages }: {
         </Button>
 
         {/* Voice control */}
-        <VoicesControl availableVoices={availableVoices} setVoiceAndRestart={setVoiceAndRestart} />
+        <VoicesControl availableVoices={availableVoices} setVoiceAndRestart={setVoiceAndRestart} voiceApiFailed={voiceApiFailed} />
       </div>
     </div>
   );

src/components/player/VoicesControl.tsx

@@ -1,5 +1,6 @@
 'use client';
 
+import { useState } from 'react';
 import {
   Listbox,
   ListboxButton,
@@ -9,15 +10,45 @@ import {
 import { ChevronUpDownIcon } from '@/components/icons/Icons';
 import { useConfig } from '@/contexts/ConfigContext';
 
-export const VoicesControl = ({ availableVoices, setVoiceAndRestart }: {
+export const VoicesControl = ({ availableVoices, setVoiceAndRestart, voiceApiFailed }: {
   availableVoices: string[];
   setVoiceAndRestart: (voice: string) => void;
+  voiceApiFailed: boolean;
 }) => {
   const { voice: configVoice } = useConfig();
+  const [customVoice, setCustomVoice] = useState(configVoice);
 
   // Use configVoice as the source of truth
   const currentVoice = configVoice;
 
+  // Show text input only if API failed
+  if (voiceApiFailed) {
+    return (
+      <div className="relative">
+        <input
+          type="text"
+          value={customVoice}
+          onChange={(e) => setCustomVoice(e.target.value)}
+          onKeyDown={(e) => {
+            if (e.key === 'Enter' && customVoice.trim()) {
+              setVoiceAndRestart(customVoice.trim());
+            }
+          }}
+          onBlur={() => {
+            if (customVoice.trim() && customVoice !== configVoice) {
+              setVoiceAndRestart(customVoice.trim());
+            } else {
+              setCustomVoice(configVoice);
+            }
+          }}
+          placeholder="Enter voice"
+          className="bg-transparent text-foreground text-xs sm:text-sm focus:outline-none border border-accent rounded px-1.5 sm:px-2 py-0.5 sm:py-1 w-24 sm:w-28"
+          title="Voice API unavailable - enter custom voice"
+        />
+      </div>
+    );
+  }
+
   return (
     <div className="relative">
       <Listbox value={currentVoice} onChange={setVoiceAndRestart}>

src/contexts/TTSContext.tsx

@@ -63,6 +63,7 @@ interface TTSContextType {
 
   // Voice settings
   availableVoices: string[];
+  voiceApiFailed: boolean;
 
   // Control functions
   togglePlay: () => void;
@@ -110,7 +111,7 @@ export function TTSProvider({ children }: { children: ReactNode }): ReactElement
   // Remove OpenAI client reference as it's no longer needed
   const audioContext = useAudioContext();
   const audioCache = useAudioCache(25);
-  const { availableVoices, fetchVoices } = useVoiceManagement(openApiKey, openApiBaseUrl);
+  const { availableVoices, fetchVoices, voiceApiFailed } = useVoiceManagement(openApiKey, openApiBaseUrl);
 
   // Add ref for location change handler
   const locationChangeHandlerRef = useRef<((location: string | number) => void) | null>(null);
@@ -868,6 +869,7 @@ export function TTSProvider({ children }: { children: ReactNode }): ReactElement
     currDocPageNumber,
     currDocPages,
     availableVoices,
+    voiceApiFailed,
     togglePlay,
     skipForward,
     skipBackward,
@@ -892,6 +894,7 @@ export function TTSProvider({ children }: { children: ReactNode }): ReactElement
     currDocPageNumber,
     currDocPages,
     availableVoices,
+    voiceApiFailed,
     togglePlay,
     skipForward,
     skipBackward,

src/hooks/audio/useVoiceManagement.ts

@@ -12,6 +12,7 @@ const DEFAULT_VOICES = ['alloy', 'ash', 'coral', 'echo', 'fable', 'onyx', 'nova'
  */
 export function useVoiceManagement(apiKey: string | undefined, baseUrl: string | undefined) {
   const [availableVoices, setAvailableVoices] = useState<string[]>([]);
+  const [voiceApiFailed, setVoiceApiFailed] = useState(false);
 
   const fetchVoices = useCallback(async () => {
     try {
@@ -23,16 +24,18 @@ export function useVoiceManagement(apiKey: string | undefined, baseUrl: string |
           'Content-Type': 'application/json',
         },
       });
-      
+
       if (!response.ok) throw new Error('Failed to fetch voices');
       const data = await response.json();
       setAvailableVoices(data.voices || DEFAULT_VOICES);
+      setVoiceApiFailed(data.failed || false);
     } catch (error) {
       console.error('Error fetching voices:', error);
       // Set available voices to default openai voices
       setAvailableVoices(DEFAULT_VOICES);
+      setVoiceApiFailed(true);
     }
   }, [apiKey, baseUrl]);
 
-  return { availableVoices, fetchVoices };
+  return { availableVoices, fetchVoices, voiceApiFailed };
 }