Remove speechMarks write stream

packages/text-to-speech/src/index.ts

@@ -90,6 +90,7 @@ export const textToSpeechHandler = Sentry.GCPFunction.wrapHttpFunction(
       console.error(e)
       return res.status(200).send('UNAUTHENTICATED')
     }
+    // validate input
     const input = req.body as HTMLInput
     const id = input.id
     const bucket = input.bucket
@@ -97,26 +98,30 @@ export const textToSpeechHandler = Sentry.GCPFunction.wrapHttpFunction(
       return res.status(200).send('Invalid data')
     }
     try {
+      // audio file to be saved in GCS
       const audioFileName = `speech/${id}.mp3`
       const audioFile = createGCSFile(bucket, audioFileName)
       const audioStream = audioFile.createWriteStream({
         resumable: true,
       }) as NodeJS.WriteStream
-      const speechMarksFileName = `speech/${id}.json`
-      const speechMarksFile = createGCSFile(bucket, speechMarksFileName)
-      const speechMarksStream = speechMarksFile.createWriteStream({
-        resumable: true,
-      }) as NodeJS.WriteStream
+      // synthesize text to speech
       const startTime = Date.now()
-      await synthesizeTextToSpeech({
+      const { speechMarks } = await synthesizeTextToSpeech({
         ...input,
         textType: 'html',
         audioStream,
-        speechMarksStream,
       })
       console.info(
         `Synthesize text to speech completed in ${Date.now() - startTime} ms`
       )
+      // speech marks file to be saved in GCS
+      const speechMarksFileName = `speech/${id}.json`
+      await uploadToBucket(
+        speechMarksFileName,
+        Buffer.from(JSON.stringify(speechMarks)),
+        bucket
+      )
+      // update speech state
       const updated = await updateSpeech(
         id,
         token,
@@ -124,12 +129,10 @@ export const textToSpeechHandler = Sentry.GCPFunction.wrapHttpFunction(
         audioFileName,
         speechMarksFileName
       )
-
       if (!updated) {
         console.error('Failed to update speech')
         return res.status(500).send({ errorCodes: 'DB_ERROR' })
       }
-
       console.info('Text to speech cloud function completed')
       res.send('OK')
     } catch (e) {

packages/text-to-speech/src/textToSpeech.ts

@@ -18,7 +18,6 @@ export interface TextToSpeechInput {
   rate?: number
   complimentaryVoice?: string
   audioStream?: NodeJS.ReadWriteStream
-  speechMarksStream?: NodeJS.ReadWriteStream
 }
 
 export interface TextToSpeechOutput {
@@ -42,7 +41,6 @@ export const synthesizeTextToSpeech = async (
   }
   const textType = input.textType || 'html'
   const audioStream = input.audioStream || new PassThrough()
-  const speechMarksStream = input.speechMarksStream || new PassThrough()
   const speechConfig = SpeechConfig.fromSubscription(
     process.env.AZURE_SPEECH_KEY,
     process.env.AZURE_SPEECH_REGION
@@ -93,15 +91,13 @@ export const synthesizeTextToSpeech = async (
         e.text
       }`
     )
-    const speechMark: SpeechMark = {
+    speechMarks.push({
       word: e.text,
       time: (timeOffset + e.audioOffset) / 10000,
       start: e.textOffset,
       length: e.wordLength,
       type: 'word',
-    }
-    speechMarks.push(speechMark)
-    speechMarksStream.write(Buffer.from(JSON.stringify(speechMark)))
+    })
   }
 
   synthesizer.bookmarkReached = (s, e) => {
@@ -110,13 +106,11 @@ export const synthesizeTextToSpeech = async (
         e.audioOffset / 10000
       }ms, bookmark text: ${e.text}`
     )
-    const speechMark: SpeechMark = {
+    speechMarks.push({
       word: e.text,
       time: (timeOffset + e.audioOffset) / 10000,
       type: 'bookmark',
-    }
-    speechMarks.push(speechMark)
-    speechMarksStream.write(Buffer.from(JSON.stringify(speechMark)))
+    })
   }
 
   const speakSsmlAsyncPromise = (
@@ -150,6 +144,7 @@ export const synthesizeTextToSpeech = async (
         timeOffset = timeOffset + result.audioDuration
       }
     } else {
+      // assemble ssml
       const ssml = `${startSsml(null, ssmlOptions)}${input.text}${endSsml()}`
       await speakSsmlAsyncPromise(ssml)
     }
@@ -159,7 +154,6 @@ export const synthesizeTextToSpeech = async (
   } finally {
     console.debug('closing synthesizer')
     audioStream.end()
-    speechMarksStream.end()
     synthesizer.close()
     console.debug('synthesizer closed')
   }