Variable Gap between playing queued streams

[yosun]

Feature Request

Variable Gap between playing queued streams

Is your feature request related to a problem? Please describe.

Current queue seems to lead to distortion of voices.

Describe the solution you'd like

Maybe adding in 0.2f seconds or variable time between each could help.

[StephenHodgson]

You should be able to handle the queue playback in your implementation.

The API will send you clips as soon as they're loaded. Which could be faster than the playback speed.

[StephenHodgson]

See the example implementation in the samples scene

[StephenHodgson]

https://github.com/RageAgainstThePixel/com.rest.elevenlabs/blob/3ee6fdd5395d4de1a31a0bb0d7b88288c7a1337d/ElevenLabs/Packages/com.rest.elevenlabs/Samples~/TextToSpeech/TextToSpeechDemo.cs#L68-L76

[yosun]

private AudioSource audioSource;
private Queue<AudioClip> streamClipQueue = new Queue<AudioClip>();

private bool isPlaying = false;
private bool isTransitioning = false;

private void Start()
{
    client = new ElevenLabsClient(APIKEY_11);
    audioSource = GetComponent<AudioSource>(); // Assign the AudioSource here

    if (!string.IsNullOrEmpty(defaultVoiceId))
    {
        SetVoice(defaultVoiceId);
    }

    lifetimeCancellationTokenSource = new CancellationTokenSource();

    StartCoroutine(PlayStreamedClips());
}

private IEnumerator PlayStreamedClips()
{
    while (true)
    {
        if (!isPlaying && streamClipQueue.Count > 0)
        {
            var nextClip = streamClipQueue.Dequeue();
            Debug.Log($"Playing {nextClip.name}");

            if (audioSource.isPlaying)
            {
                isTransitioning = true;
                yield return new WaitForSeconds(0.1f); // Adjust this overlap time for smoother transition
                isTransitioning = false;
            }

            isPlaying = true;
            audioSource.clip = nextClip;
            audioSource.Play();

            yield return new WaitUntil(() => !audioSource.isPlaying || isTransitioning);
            isPlaying = false;
        }

        yield return null;
    }
}

[yosun]

hey.... actually this is a bug - it seems that your dequeue system for short words will swap the order. your eleven labs call should check for text queue order before playing next queued

https://github.com/RageAgainstThePixel/com.rest.elevenlabs/issues/46