mcspeak/docker-compose.yml
Ryan Malloy 538b8a513e Replace Ollama with llama-server for 15x Orpheus throughput
Build llama.cpp from source with SM 120 CUDA kernels and FORCE_CUBLAS
for RTX 5070 Blackwell. Rewrite OrpheusEngine to stream tokens via SSE
and decode SNAC in overlapping 28-token batches (4 frames), replacing
the blocking requests+stream:false approach.

Performance: 13.5 → 170-213 tok/s. 100s audio generates in ~48s (2x
faster than realtime). Replaces requests with httpx async client.

Also switch MCP transport to stateless_http mode so container restarts
don't invalidate client sessions.
2026-02-21 21:33:23 -07:00

68 lines
1.8 KiB
YAML

services:
tts-mcp:
build: .
container_name: tts-mcp
restart: unless-stopped
env_file: .env
environment:
# Override for Docker networking (container DNS instead of IPs)
TTS_PIPER_HOST: piper-tts
TTS_ORPHEUS_URL: http://llama-server:8081
# PipeWire client config
XDG_RUNTIME_DIR: /run/user/1000
volumes:
# Kokoro ONNX models (read-only)
- ./models:/app/models:ro
# HuggingFace cache for SNAC model download (lazy-loaded on first Orpheus call)
- hf-cache:/home/tts/.cache/huggingface
# PipeWire socket for audio playback through host speakers
- /run/user/1000/pipewire-0:/run/user/1000/pipewire-0
depends_on:
llama-server:
condition: service_healthy
networks:
- caddy
- dootie-internal
labels:
caddy: voice.l.supported.systems
caddy.reverse_proxy: "{{upstreams 8371}}"
llama-server:
build:
context: .
dockerfile: llama-server.Dockerfile
container_name: orpheus-llama-server
restart: unless-stopped
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: 1
capabilities: [gpu]
volumes:
# GGUF model file (set ORPHEUS_GGUF_PATH in .env, e.g. from Ollama blob storage)
- ${ORPHEUS_GGUF_PATH}:/models/orpheus.gguf:ro
command: >-
--host 0.0.0.0 --port 8081
--model /models/orpheus.gguf
--n-gpu-layers 999 --ctx-size 4096
--flash-attn --cont-batching
networks:
- dootie-internal
healthcheck:
test: ["CMD", "curl", "-sf", "http://127.0.0.1:8081/health"]
interval: 15s
timeout: 5s
start_period: 120s
retries: 5
volumes:
hf-cache:
networks:
caddy:
external: true
dootie-internal:
external: true