mARTin-B78
92e2381d0b
- run_server.py: warm up CUDA graphs at server startup via lifespan event so the first real request does not pay the 7-8s graph- compilation penalty; uses modern lifespan API instead of deprecated on_event - generate_voices.py: set chunk_size=4 per voice so streaming clients receive first audio after ~333ms instead of ~1s - docker-compose.yml: add --max-seq-len 2048 (halves static KV cache, reduces VRAM and graph-capture time) - config/benchmark_api.py: new script to measure TTFA, RTF and speed against the live API endpoint Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
42 lines
1.1 KiB
YAML
42 lines
1.1 KiB
YAML
services:
|
|
faster-qwen3-tts:
|
|
image: martinb78/faster-qwen3-tts-dgx-spark:latest
|
|
container_name: faster-qwen3-tts
|
|
restart: unless-stopped
|
|
runtime: nvidia
|
|
environment:
|
|
- NVIDIA_VISIBLE_DEVICES=all
|
|
# Optional: set HF_TOKEN if model needs downloading from HuggingFace
|
|
# - HF_TOKEN=${HF_TOKEN}
|
|
ports:
|
|
- "8020:8000"
|
|
volumes:
|
|
# Mount your local Qwen3-TTS model (download first, see README)
|
|
- ${MODEL_PATH:-./models/Qwen3-TTS}:/models/Qwen3-TTS:ro
|
|
# Config directory (voices.json generated on startup)
|
|
- ./config:/config:rw
|
|
# Speaker reference audio files
|
|
- ./config/speakers:/voices:ro
|
|
command: >
|
|
/bin/bash -c "
|
|
python3 /config/generate_voices.py &&
|
|
python3 /config/run_server.py
|
|
--model /models/Qwen3-TTS
|
|
--voices /config/voices.json
|
|
--port 8000
|
|
--max-seq-len 2048
|
|
"
|
|
deploy:
|
|
resources:
|
|
reservations:
|
|
devices:
|
|
- driver: nvidia
|
|
count: all
|
|
capabilities: [gpu]
|
|
networks:
|
|
- dgx_net
|
|
|
|
networks:
|
|
dgx_net:
|
|
external: true
|