Decibri is a cross-platform audio engine for real-time speech and voice applications. One unified API across Python, Rust, Node.js, and browsers.
Key features:
- Microphone capture and speaker playback
- Voice activity detection (Silero or energy modes)
- Async and streaming APIs
- Cross-platform: Windows, macOS, Linux, browsers
- Available in Python, Node.js, and Rust
- Cloud and local speech-to-text integrations
To learn more, visit decibri.com/docs/.
Python (recommended with uv):
uv pip install decibriOr with pip:
pip install decibriNode.js:
npm install decibriRust:
cargo add decibriimport decibri
with decibri.Microphone(sample_rate=16000) as mic:
for chunk in mic:
print(f"Got {len(chunk)} bytes")
breakFull Python guide: here.
const { Microphone } = require('decibri');
const mic = new Microphone({ sampleRate: 16000 });
mic.on('data', (chunk) => console.log(`Got ${chunk.length} bytes`));
setTimeout(() => mic.stop(), 5000);Full Node.js and browser guide: here.
use decibri::{Microphone, MicrophoneConfig};
fn main() -> Result<(), Box<dyn std::error::Error>> {
let microphone = Microphone::new(MicrophoneConfig::default())?;
let stream = microphone.start()?;
while let Ok(Some(chunk)) = stream.next_chunk(None) {
println!("Got {} bytes", chunk.data.len());
}
Ok(())
}Full Rust guide: here.
Supports:
- Windows x64
- macOS arm64
- Linux x64
- Linux arm64
Decibri ships two VAD modes:
- Energy mode: lightweight RMS threshold.
- Silero mode: ML-based detection using the Silero VAD v5 ONNX model. More accurate, especially in noisy environments. The model (~2.3 MB) is bundled; no downloads or API keys required.
import decibri
with decibri.Microphone(sample_rate=16000, vad="silero", vad_threshold=0.5) as mic:
for chunk in mic:
print(f"VAD score {mic.vad_score}; speaking={mic.is_speaking}")
breakThe same VAD API is available in Node.js and Rust. Language-specific examples and tuning advice live in each binding's README.
Decibri pipes microphone audio directly into cloud and local speech services. Full guides at decibri.com/docs/integrations.
| Provider | Environment | STT | TTS | VAD | KWS |
|---|---|---|---|---|---|
| OpenAI Realtime | Cloud | ✅ | ✅ | ||
| Deepgram | Cloud | ✅ | ✅ | ||
| AssemblyAI | Cloud | ✅ | |||
| Mistral Voxtral | Cloud | ✅ | |||
| AWS Transcribe | Cloud | ✅ | |||
| Google Speech | Cloud | ✅ | ✅ | ||
| Azure Speech | Cloud | ✅ | ✅ | ||
| Sherpa-ONNX | Local | ✅ | ✅ | ✅ | ✅ |
| Whisper.cpp | Local | ✅ |
Decibri is built on a single Rust core. The same audio engine powers all three language bindings, ensuring consistent behavior regardless of which language you use.
Python and Node.js install pre-built native binaries for your platform. Browsers run the same API via a JavaScript implementation that ships inside the npm package.
Issues and PRs welcome. See CONTRIBUTING.md for development setup.
Apache-2.0 © 2026 Decibri.