shitton

src/ai/speech-to-text.ts

@@ -198,7 +198,7 @@ async function transcribeGeminiSpeech(audio: AiDownloadedFile, signal?: AbortSig
             temperature: 0,
             abortSignal: signal,
         },
-    });
+    }) as unknown as GeminiSpeechResponse;
 
     return {
         provider: AiProvider.GEMINI,
@@ -240,17 +240,19 @@ async function transcribeOllamaSpeech(audio: AiDownloadedFile, signal?: AbortSig
     };
 }
 
-function collectGeminiText(response: any): string {
-    if (typeof response?.text === "string") return response.text;
+type GeminiSpeechResponse = {
+    text?: string;
+    candidates?: Array<{content?: {parts?: Array<{text?: string}>}}> ;
+};
 
-    const candidates = response?.candidates ?? [];
-    const candidateText = candidates
-        .flatMap((candidate: any) => candidate?.content?.parts ?? [])
-        .map((part: any) => part?.text ?? "")
+function collectGeminiText(response: GeminiSpeechResponse): string {
+    if (typeof response.text === "string") return response.text;
+
+    const candidateText = (response.candidates ?? [])
+        .flatMap(candidate => candidate.content?.parts ?? [])
+        .map(part => part.text ?? "")
         .join("");
     if (candidateText.trim()) return candidateText;
 
-    return (response?.candidates ?? [])
-        .map((output: any) => typeof output === "string" ? output : output?.content?.parts?.[0]?.text ?? "")
-        .join("");
+    return "";
 }