Noise Issue with TTS Conversion to Base64

API
1

I’m experiencing an issue with my code for TTS generation. While the TTS output itself seems fine when I check it, the resulting audio after processing ends up as noise. I suspect the problem lies in the conversion of the TTS output to Base64 format:

async generate(gptReply, interactionCount) {
    const { partialResponseIndex, partialResponse } = gptReply;

    if (!partialResponse) { return; }

    try {
        const response = await fetch(
            'https://api.openai.com/v1/audio/speech',
            {
                method: 'POST',
                headers: {
                    'Authorization': `Bearer ${process.env.OPENAI_API_KEY}`,
                    'Content-Type': 'application/json',
                },
                body: JSON.stringify({
                    input: partialResponse,
                    model: 'tts-1',
                    voice: 'alloy',
                    response_format: 'opus',
                }),
            }
        );

        if (response.status === 200) {
            try {
                const audioArrayBuffer = await response.arrayBuffer();
                const base64String = Buffer.from(audioArrayBuffer).toString('base64');
                this.emit('speech', partialResponseIndex, base64String, partialResponse, interactionCount);
            } catch (err) {
                console.error('Error converting audio to Base64:', err);
            }
        } else {
            console.error('OpenAI TTS error:', response);
        }
    } catch (err) {
        console.error('Error occurred in TextToSpeech service:', err);
    }
}

I would appreciate any advice or insights that can help resolve this issue.

2

Hi,

Did you set the right parameters of the codec for conversion to base 64?
If I remember correctly (only of the top off my head rn) I had a similar issue when converting due to the codec: raw 16 bit PCM audio at 24kHz, 1 channel, little-endian.

Let me know if this helps. If not, I can go and check in my code.

3

Thx, I tried it but couldn’t get it to work. If possible, I’d appreciate it if you could check your code.

4

You’re requesting a file in opus format.

Save it binary.

Name it speech.ogg