feat: support non-english languages

experiment-station · Dec 16, 2023 · a65b25c · a65b25c
1 parent 0e2f2d9
commit a65b25c
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 6 deletions.
diff --git a/lib/services/ai/deepgram.ts b/lib/services/ai/deepgram.ts
@@ -4,13 +4,20 @@ import { z } from 'zod';
 
 const deepgram = createDeepgramClient(env.DEEPGRAM_API_KEY);
 
-export const transcribeAudio = async ({ fileURL }: { fileURL: string }) => {
+export const transcribeAudio = async ({
+  language = 'en',
+  url,
+}: {
+  language?: string;
+  url: string;
+}) => {
   const { error, result } = await deepgram.listen.prerecorded.transcribeUrl(
     {
-      url: fileURL,
+      url,
     },
     {
-      model: 'nova-2',
+      language,
+      model: language === 'en' ? 'nova-2' : 'base',
       paragraphs: true,
       smart_format: true,
     },

diff --git a/lib/services/ai/transcribe-episode.ts b/lib/services/ai/transcribe-episode.ts
@@ -28,16 +28,19 @@ export const transcribeEpisode = async (id: Tables<'episode'>['id']) => {
 
   const episodeQuery = await supabase
     .from('episode')
-    .select('audio_url')
+    .select('audio_url, show(language)')
     .eq('id', id)
     .single();
 
   if (episodeQuery.error) {
     throw new DatabaseError(episodeQuery.error);
   }
 
-  const fileURL = await getFinalRedirectURL(episodeQuery.data.audio_url);
-  const transcription = await transcribeAudio({ fileURL });
+  const url = await getFinalRedirectURL(episodeQuery.data.audio_url);
+  const transcription = await transcribeAudio({
+    language: episodeQuery.data.show?.language ?? undefined,
+    url,
+  });
 
   const updateEpisodeContentQuery = await supabase
     .from('episode_content')