discord.js - Imagine an app•9mo ago

Problem saving audioBuffer to file - whisper api

Hi, how can i convert the audioBuffer into a e.g. wav file? I want to use whisper for spech regognition. this is my code:

const tmpFile = await new Promise((resolve, reject) => {
    tmp.file({ postfix: ".ogg" }, (err, path, fd, cleanup) => {
      if (err) {
        reject(err);
        return;
      }

      fs.writeFile(path, msg.audioBuffer, (err) => {
        if (err) {
          reject(err);
          return;
        }

        resolve({ path, cleanup });
      });
    });
  });

const tmpFile = await new Promise((resolve, reject) => {
    tmp.file({ postfix: ".ogg" }, (err, path, fd, cleanup) => {
      if (err) {
        reject(err);
        return;
      }

      fs.writeFile(path, msg.audioBuffer, (err) => {
        if (err) {
          reject(err);
          return;
        }

        resolve({ path, cleanup });
      });
    });
  });

The file is written successfully, but I can't play it (unrecognized file format) I also tried to post audioBuffer directly, but that doesn't work either.

async function resolveSpeechWithWhisper(audioBuffer) {
  const transcription = await openai.audio.transcriptions.create({
    file: audioBuffer,
    model: "whisper-1",
    response_format: "verbose_json",
    timestamp_granularities: ["word"],
  });
}

async function resolveSpeechWithWhisper(audioBuffer) {
  const transcription = await openai.audio.transcriptions.create({
    file: audioBuffer,
    model: "whisper-1",
    response_format: "verbose_json",
    timestamp_granularities: ["word"],
  });
}

error: BadRequestError: 400 Could not parse multipart form at APIError.generate (/home/git/discord-speech-recognition/examples/simpleBot/node_modules/openai/error.js:44:20) at OpenAI.makeStatusError (/home/git/discord-speech-recognition/examples/simpleBot/node_modules/openai/core.js:263:33) at OpenAI.makeRequest (/home/git/discord-speech-recognition/examples/simpleBot/node_modules/openai/core.js:306:30) at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at async Object.resolveSpeechWithWhisper [as speechRecognition] (/home/git/discord-speech-recognition/examples/simpleBot/index.js:20:25) {

"@discordjs/opus": "^0.8.0",
    "@discordjs/voice": "^0.11.0",
    "discord-speech-recognition": "^3.4.1",
    "discord.js": "^14.3.0",
    "dotenv": "^16.4.5",
    "openai": "^4.29.2",
    "tmp": "^0.2.3",
    "tweetnacl": "^1.0.3"

"@discordjs/opus": "^0.8.0",
    "@discordjs/voice": "^0.11.0",
    "discord-speech-recognition": "^3.4.1",
    "discord.js": "^14.3.0",
    "dotenv": "^16.4.5",
    "openai": "^4.29.2",
    "tmp": "^0.2.3",
    "tweetnacl": "^1.0.3"

Can anyone help me with this? Thanks

5 Replies

d.js toolkit•9mo ago

- What's your exact discord.js npm list discord.js and node node -v version? - Not a discord.js issue? Check out #other-js-ts. - Consider reading #how-to-get-help to improve your question! - Explain what exactly your issue is. - Post the full error stack trace, not just the top part! - Show your code! - Issue solved? Press the button! - ✅ Marked as resolved by OP

jaemilOP•9mo ago

node: v20.11.0 im using the SpeechEvents.speech event.

client.on(SpeechEvents.speech, async (msg) => {
  // If bot didn't recognize speech, content will be empty
  // if (!msg.content) return;

  console.log("msg", msg);

  console.log("audioBuffer type:", typeof msg.audioBuffer);
  console.log("audioBuffer:", msg.audioBuffer);
}

client.on(SpeechEvents.speech, async (msg) => {
  // If bot didn't recognize speech, content will be empty
  // if (!msg.content) return;

  console.log("msg", msg);

  console.log("audioBuffer type:", typeof msg.audioBuffer);
  console.log("audioBuffer:", msg.audioBuffer);
}

const client = new Client({
  intents: [
    GatewayIntentBits.GuildVoiceStates,
    GatewayIntentBits.GuildMessages,
    GatewayIntentBits.Guilds,
    GatewayIntentBits.MessageContent,
  ],
});
addSpeechEvent(client, {
  speechRecognition: resolveSpeechWithWhisper,
  ignoreBots: true,
});

const client = new Client({
  intents: [
    GatewayIntentBits.GuildVoiceStates,
    GatewayIntentBits.GuildMessages,
    GatewayIntentBits.Guilds,
    GatewayIntentBits.MessageContent,
  ],
});
addSpeechEvent(client, {
  speechRecognition: resolveSpeechWithWhisper,
  ignoreBots: true,
});

jaemilOP•9mo ago

im using this repo https://github.com/Rei-x/discord-speech-recognition

GitHub

GitHub - Rei-x/discord-speech-recognition: Speech to text extension...

Speech to text extension for discord.js. Contribute to Rei-x/discord-speech-recognition development by creating an account on GitHub.

jaemilOP•9mo ago

const opusStream = receiver.subscribe(userId, {
        end: {
          behavior: EndBehaviorType.AfterSilence,
          duration: 300,
        },
      });

      const bufferData: Uint8Array[] = [];
      opusStream
        .pipe(new OpusDecodingStream())
        .on("data", (data: Uint8Array) => {
          bufferData.push(data);
        });

      opusStream.on("end", async () => {
        const voiceMessage = await createVoiceMessage({
          client,
          bufferData,
          user,
          connection,
          speechOptions,
        });

        if (voiceMessage) client.emit(SpeechEvents.speech, voiceMessage);
      });

const opusStream = receiver.subscribe(userId, {
        end: {
          behavior: EndBehaviorType.AfterSilence,
          duration: 300,
        },
      });

      const bufferData: Uint8Array[] = [];
      opusStream
        .pipe(new OpusDecodingStream())
        .on("data", (data: Uint8Array) => {
          bufferData.push(data);
        });

      opusStream.on("end", async () => {
        const voiceMessage = await createVoiceMessage({
          client,
          bufferData,
          user,
          connection,
          speechOptions,
        });

        if (voiceMessage) client.emit(SpeechEvents.speech, voiceMessage);
      });

this is how audioBuffer is created https://github.com/Rei-x/discord-speech-recognition/blob/main/src/bot/events/speech.ts

GitHub

discord-speech-recognition/src/bot/events/speech.ts at main · Rei-x...

Speech to text extension for discord.js. Contribute to Rei-x/discord-speech-recognition development by creating an account on GitHub.

jaemilOP•9mo ago

Thank you so much, it did work. now i only need to double the sample rate from 48000 to 96000. idk why the recorded audio is in slowmo..

Gaming

Programming

Problem saving audioBuffer to file - whisper api