Willow Inference Server Skill

Local ASR (speech-to-text) and TTS (text-to-speech) inference server.

Setup

1. Start Willow Inference Server

git clone https://github.com/toverainc/willow-inference-server.git
cd willow-inference-server
./utils.sh install
./utils.sh gen-cert your-hostname
./utils.sh run

Server runs at https://your-hostname:19000

2. Configure Environment

Set the server URL:

export WILLOW_BASE_URL="https://your-hostname:19000"

Or configure per request (see below).

ASR (Speech-to-Text)

Transcribe Audio File

curl -X POST "${WILLOW_BASE_URL}/api/asr" \
  -F "audio_file=@/path/to/audio.m4a" \
  -F "language=auto"

Parameters

Supported Formats

• MP3, WAV, M4A, OGG, FLAC, WebM

Example: Transcribe with curl

# Basic transcription
curl -X POST "${WILLOW_BASE_URL}/asr" \
  -F "audio_file=@recording.m4a" \
  -F "language=zh"

# With specific model
curl -X POST "${WILLOW_BASE_URL}/asr" \
  -F "audio_file=@meeting.mp3" \
  -F "language=en" \
  -F "model=base"

TTS (Text-to-Speech)

Convert Text to Speech

curl -X POST "${WILLOW_BASE_URL}/tts" \
  -H "Content-Type: application/json" \
  -d '{"text": "Hello world", "voice": "af_sarah"}'

Parameters

Available Voices

Common voices (format: gender_voicename):

• af_sarah - Sarah (Female)
• af_bella - Bella (Female)
• am_michael - Michael (Male)
• am_alex - Alex (Male)

Check server docs for full list: ${WILLOW_BASE_URL}/api/docs

Example: TTS with curl

# Basic TTS
curl -X POST "${WILLOW_BASE_URL}/tts" \
  -H "Content-Type: application/json" \
  -d '{"text": "你好，这是测试"}' \
  -o output.wav

# With custom voice
curl -X POST "${WILLOW_BASE_URL}/tts" \
  -H "Content-Type: application/json" \
  -d '{"text": "Hello!", "voice": "am_michael", "speed": 1.2}' \
  -o hello.mp3

Environment Variables

Workflow Examples

1. Record and Transcribe

# Record audio (macOS)
rec test.wav

# Transcribe
curl -X POST "${WILLOW_BASE_URL}/asr" \
  -F "audio_file=@test.wav" \
  -F "language=auto"

2. Text to Speech

# Convert text to speech
curl -X POST "${WILLOW_BASE_URL}/tts" \
  -H "Content-Type: application/json" \
  -d '{"text": "今天的任务是学习新技能"}' \
  -o speech.wav

3. Batch Transcription

for f in *.m4a; do
  curl -X POST "${WILLOW_BASE_URL}/asr" \
    -F "audio_file=@$f" \
    -F "language=auto" \
    -o "${f%.m4a}.txt"
done

API Documentation

Full API docs available at: ${WILLOW_BASE_URL}/api/docs

Notes

• All endpoints require HTTPS (or HTTP if configured)
• Audio files are processed locally on the server
• ASR latency depends on model size and hardware
• TTS voices can be customized with custom voice recordings