c86e323d7a
Quality-first multi-pass audiobook generation from English manuscripts. Includes: - umbraaudio/ core package (pipeline, config, voices, web review UI) - scripts/ CLI utilities + model download script - tts_server/ remote TTS server - specs/ design docs, tests/, pyproject.toml + uv.lock - Qwen3-TTS git submodule (pinned 022e286) with local demo.py UI patch saved at scripts/qwen3_tts_demo_umbra_patch.patch - .env with runtime config/secrets Excluded (see .gitignore): models/ (downloaded via scripts/download_models.sh), output/, logs/, .venv, .umbra runtime data, and book projects (DealHunterShip_*), which are committed separately.
34 lines
1.1 KiB
Python
34 lines
1.1 KiB
Python
import os
|
|
import torch
|
|
import soundfile as sf
|
|
from qwen_tts import Qwen3TTSModel
|
|
|
|
model_path = "/Volumes/1TB_APL_Storage/UmbraAudio/models/Qwen3-TTS-12Hz-1.7B-Base"
|
|
out_dir = "/Volumes/1TB_APL_Storage/UmbraAudio/output"
|
|
os.makedirs(out_dir, exist_ok=True)
|
|
|
|
device = os.environ.get("DEVICE", "mps")
|
|
dtype = torch.bfloat16 if device != "cpu" else torch.float32
|
|
|
|
print(f"Loading model on {device} (dtype={dtype}) ...")
|
|
model = Qwen3TTSModel.from_pretrained(
|
|
model_path,
|
|
device_map=device,
|
|
dtype=dtype,
|
|
attn_implementation="eager",
|
|
)
|
|
|
|
ref_audio = "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-TTS-Repo/clone.wav"
|
|
ref_text = "Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! And thanks to you."
|
|
|
|
print("Generating voice clone ...")
|
|
wavs, sr = model.generate_voice_clone(
|
|
text="This is a local test of Qwen three TTS running on my Mac mini. The audio should sound natural and clear.",
|
|
language="English",
|
|
ref_audio=ref_audio,
|
|
ref_text=ref_text,
|
|
)
|
|
out = os.path.join(out_dir, "test_voice_clone.wav")
|
|
sf.write(out, wavs[0], sr)
|
|
print(f"DONE: wrote {out} ({sr} Hz, {len(wavs[0])/sr:.2f}s)")
|