DeepFilterNet3 — Android
DeepFilterNet3 runs on Android and embedded Linux via ONNX Runtime, using STFT/ERB-based noise suppression. It processes microphone input before VAD and STT, providing cleaner transcription in noisy environments.
Model
| Model | Backend | Size | HuggingFace |
|---|---|---|---|
| DeepFilterNet3 | ONNX Runtime | ~8 MB | aufklarer/DeepFilterNet3-ONNX |
Performance
| Metric | Value |
|---|---|
| RTF | ~0.15 (real-time capable) |
| Processing | STFT/ERB-based noise suppression |
| Model file | deepfilter-auxiliary.bin |
Important
On Android, noise cancellation is integrated into the SpeechPipeline and runs automatically on mic input before VAD and STT. No standalone API is needed. See speech-android on GitHub for integration details.
Source code: github.com/soniqo/speech-android