Ryan Malloy
538b8a513e
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.
37 lines
762 B
TOML
37 lines
762 B
TOML
[build-system]
|
|
requires = ["hatchling"]
|
|
build-backend = "hatchling.build"
|
|
|
|
[project]
|
|
name = "tts-mcp"
|
|
version = "2026.02.20"
|
|
description = "Multi-engine TTS server with speech queue, exposed via FastMCP 3.0 Streamable HTTP"
|
|
requires-python = ">=3.12"
|
|
license = "MIT"
|
|
authors = [{name = "Ryan Malloy", email = "ryan@supported.systems"}]
|
|
dependencies = [
|
|
"fastmcp>=3.0.0",
|
|
"httpx",
|
|
"kokoro-onnx>=0.5.0",
|
|
"numpy",
|
|
"onnxruntime",
|
|
"pydantic-settings",
|
|
"snac>=1.2.1",
|
|
"soundfile",
|
|
"torch",
|
|
"wyoming>=1.8.0",
|
|
]
|
|
|
|
[project.scripts]
|
|
tts-mcp = "tts_mcp.__main__:main"
|
|
|
|
[tool.hatch.build.targets.wheel]
|
|
packages = ["src/tts_mcp"]
|
|
|
|
[tool.ruff]
|
|
line-length = 100
|
|
target-version = "py312"
|
|
|
|
[tool.ruff.lint]
|
|
select = ["E", "F", "I", "W"]
|