Real-time speech-to-text using OpenAI Whisper (faster-whisper). Features browser audio capture, WebSocket streaming, and customizable display settings. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
34 lines
939 B
Plaintext
34 lines
939 B
Plaintext
# Server settings
|
|
HOST=0.0.0.0
|
|
PORT=5000
|
|
DEBUG=false
|
|
|
|
# Whisper settings
|
|
WHISPER_MODEL=base
|
|
# Device: cpu or cuda (for NVIDIA GPU)
|
|
WHISPER_DEVICE=cpu
|
|
# Compute type:
|
|
# CPU: int8 (fastest), float32
|
|
# GPU: float16 (recommended), int8_float16, float32
|
|
WHISPER_COMPUTE_TYPE=int8
|
|
|
|
# Audio settings
|
|
AUDIO_CHUNK_DURATION=3
|
|
AUDIO_SAMPLE_RATE=16000
|
|
|
|
# Database
|
|
DATABASE_PATH=data/settings.db
|
|
|
|
# =============================================================================
|
|
# GPU Configuration (optional)
|
|
# =============================================================================
|
|
# To enable NVIDIA GPU support:
|
|
# 1. Install NVIDIA Container Toolkit (see CLAUDE.md for instructions)
|
|
# 2. Set WHISPER_DEVICE=cuda
|
|
# 3. Set WHISPER_COMPUTE_TYPE=float16 (recommended for GPU)
|
|
# 4. Run with: docker compose -f docker-compose.yml -f docker-compose.gpu.yml up --build
|
|
#
|
|
# Example GPU settings:
|
|
# WHISPER_DEVICE=cuda
|
|
# WHISPER_COMPUTE_TYPE=float16
|