vad-non-streaming-asr-from-microphone

This example demonstrates how to use Voice Activity Detection (VAD) with non-streaming speech recognition from a microphone in Flutter with sherpa-onnx.

It uses:

Silero VAD v4 for voice activity detection
Zipformer CTC for speech recognition (default model)

Other supported ASR models: SenseVoice, Whisper. Change selectedModelIndex in lib/model_config.dart to switch models.

It works on the following platforms:

Android
iOS
Linux
macOS (both arm64 and x86_64 are supported)
Windows
Web

Features

Configurable VAD parameters (threshold, min silence, min speech, max speech)
Real-time speech detection with circle indicator (black = silent, red = speech)
Segment counter
Displays ASR transcription for each detected segment
Play back detected segments

How to build

1. Download the models

Download the VAD model and an ASR model.

Default: Zipformer CTC (Chinese)

bash

cd flutter-examples/vad-non-streaming-asr-from-microphone/assets
rm -f .gitkeep

# VAD model
curl -SL -o silero_vad.onnx https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx

# ASR model (Zipformer CTC)
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-zipformer-ctc-zh-int8-2025-07-03.tar.bz2
tar xvf sherpa-onnx-zipformer-ctc-zh-int8-2025-07-03.tar.bz2
rm sherpa-onnx-zipformer-ctc-zh-int8-2025-07-03.tar.bz2

# Remove files not required by the app (test wavs, vocab, etc.)
cd sherpa-onnx-zipformer-ctc-zh-int8-2025-07-03
rm -f *.wav README.md bbpe.model bbpe.vocab
rm -rf test_wavs
cd ..

cd ..
./generate-asset-list.py

After downloading, the assets/ directory should contain only:

assets/
├── silero_vad.onnx
└── sherpa-onnx-zipformer-ctc-zh-int8-2025-07-03/
    ├── model.int8.onnx
    └── tokens.txt

Alternative: Whisper tiny.en (English)

To use Whisper instead of Zipformer CTC, download the Whisper model and change selectedModelIndex to 2 in lib/model_config.dart:

bash

cd flutter-examples/vad-non-streaming-asr-from-microphone/assets
rm -f .gitkeep

# VAD model
curl -SL -o silero_vad.onnx https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx

# ASR model (Whisper tiny.en)
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-tiny.en.tar.bz2
tar xvf sherpa-onnx-whisper-tiny.en.tar.bz2
rm sherpa-onnx-whisper-tiny.en.tar.bz2

cd ..
./generate-asset-list.py

Then edit lib/model_config.dart and set const int selectedModelIndex = 2;.

Note: generate-asset-list.py is a symlink to ../tts/generate-asset-list.py.

2. Build the APP

For Linux

Install libmpv-dev first (required by the audioplayers package):

bash

sudo apt-get install -y libmpv-dev
flutter build linux

For macOS

bash

flutter build macos

For Windows

bash

flutter build windows

For Android

bash

flutter build apk --split-per-abi --target-platform android-arm64

For web

bash

flutter run -d chrome

For iOS

Connect your iPhone, then:

bash

flutter devices
flutter run -d <device-id> --release