From baaa38547a60be34367e3d228b47cb14027f027a Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Tue, 6 Sep 2022 17:40:48 +0800
Subject: [PATCH] Create buffer for the audio data

---
 packages/text-to-speech/src/index.ts        | 7 +++++--
 packages/text-to-speech/src/textToSpeech.ts | 9 ++++++---
 2 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/packages/text-to-speech/src/index.ts b/packages/text-to-speech/src/index.ts
index 07233a0c8..9a9c93c31 100644
--- a/packages/text-to-speech/src/index.ts
+++ b/packages/text-to-speech/src/index.ts
@@ -170,9 +170,12 @@ export const textToSpeechStreamingHandler = Sentry.GCPFunction.wrapHttpFunction(
         ...utteranceInput,
         textType: 'utterance',
       }
-      const { audioStream, speechMarks } = await synthesizeTextToSpeech(input)
+      const { audioData, speechMarks } = await synthesizeTextToSpeech(input)
+      if (!audioData) {
+        return res.status(500).send({ errorCode: 'SYNTHESIZER_ERROR' })
+      }
       res.send({
-        audioData: audioStream.read().toString('hex'),
+        audioData: audioData.toString('hex'),
         speechMarks,
       })
     } catch (e) {
diff --git a/packages/text-to-speech/src/textToSpeech.ts b/packages/text-to-speech/src/textToSpeech.ts
index db39e77b8..fa406d3b8 100644
--- a/packages/text-to-speech/src/textToSpeech.ts
+++ b/packages/text-to-speech/src/textToSpeech.ts
@@ -21,7 +21,7 @@ export interface TextToSpeechInput {
 }
 
 export interface TextToSpeechOutput {
-  audioStream: NodeJS.ReadWriteStream
+  audioData?: Buffer
   speechMarks: SpeechMark[]
 }
 
@@ -146,7 +146,11 @@ export const synthesizeTextToSpeech = async (
     } else {
       // assemble ssml
       const ssml = `${startSsml(null, ssmlOptions)}${input.text}${endSsml()}`
-      await speakSsmlAsyncPromise(ssml)
+      const result = await speakSsmlAsyncPromise(ssml)
+      return {
+        audioData: Buffer.from(result.audioData),
+        speechMarks,
+      }
     }
   } catch (error) {
     console.error('synthesis error', error)
@@ -159,7 +163,6 @@ export const synthesizeTextToSpeech = async (
   }
 
   return {
-    audioStream,
     speechMarks,
   }
 }