@vowel.to/client / index / AUDIO_CAPTURE_CONFIG
Variable: AUDIO_CAPTURE_CONFIG
ts
const AUDIO_CAPTURE_CONFIG: {
bufferSize: 2048;
};Defined in: lib/vowel/types/constants.ts:55
Audio capture chunking configuration for microphone streaming.
Controls how audio is buffered and sent to the server. Smaller chunks = more frequent sends, lower latency, better responsiveness for short phrases (e.g. "yes", "no", "okay"). Larger chunks = fewer sends, slightly less overhead.
Type Declaration
| Name | Type | Default value | Description | Defined in |
|---|---|---|---|---|
bufferSize | 2048 | 2048 | Number of samples to accumulate before sending each audio chunk. At 24kHz: 2048 samples ≈ 85ms, 4096 samples ≈ 170ms. - 2048 (default): Smaller chunks, ~2x more frequent. Matches Vowel Engine demo behavior at 48kHz input. Better for short phrases and server VAD. - 4096: Larger chunks, original behavior. Slightly less overhead. | lib/vowel/types/constants.ts:64 |
See
AUDIO_CAPTURE_CONFIG.bufferSize - Samples per chunk at output sample rate (24kHz)