Spaces:
Running
Running
Yurii Paniv
commited on
Commit
·
eb8c82c
1
Parent(s):
4fdf251
Upload app
Browse files- README.md +4 -0
- app.py +65 -0
- packages.txt +1 -0
README.md
CHANGED
|
@@ -26,3 +26,7 @@ tts-server --model_path path/to/model.pth.tar \
|
|
| 26 |
# How to train:
|
| 27 |
1. Refer to ["Nervous beginner guide"](https://tts.readthedocs.io/en/latest/tutorial_for_nervous_beginners.html) in Coqui TTS docs.
|
| 28 |
2. Instead of provided `config.json` use one from this repo.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
# How to train:
|
| 27 |
1. Refer to ["Nervous beginner guide"](https://tts.readthedocs.io/en/latest/tutorial_for_nervous_beginners.html) in Coqui TTS docs.
|
| 28 |
2. Instead of provided `config.json` use one from this repo.
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
# Attribution
|
| 32 |
+
Code for `app.py` taken from https://huggingface.co/spaces/julien-c/coqui
|
app.py
ADDED
|
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import tempfile
|
| 2 |
+
from typing import Optional
|
| 3 |
+
|
| 4 |
+
import gradio as gr
|
| 5 |
+
import numpy as np
|
| 6 |
+
|
| 7 |
+
from TTS.utils.manage import ModelManager
|
| 8 |
+
from TTS.utils.synthesizer import Synthesizer
|
| 9 |
+
|
| 10 |
+
MODEL_NAMES = [
|
| 11 |
+
"uk/mai/glow-tts"
|
| 12 |
+
]
|
| 13 |
+
MODELS = {}
|
| 14 |
+
|
| 15 |
+
manager = ModelManager()
|
| 16 |
+
|
| 17 |
+
for MODEL_NAME in MODEL_NAMES:
|
| 18 |
+
print(f"downloading {MODEL_NAME}")
|
| 19 |
+
model_path, config_path, model_item = manager.download_model(
|
| 20 |
+
f"tts_models/{MODEL_NAME}")
|
| 21 |
+
vocoder_name: Optional[str] = model_item["default_vocoder"]
|
| 22 |
+
vocoder_path = None
|
| 23 |
+
vocoder_config_path = None
|
| 24 |
+
if vocoder_name is not None:
|
| 25 |
+
vocoder_path, vocoder_config_path, _ = manager.download_model(
|
| 26 |
+
vocoder_name)
|
| 27 |
+
|
| 28 |
+
synthesizer = Synthesizer(
|
| 29 |
+
model_path, config_path, None, vocoder_path, vocoder_config_path,
|
| 30 |
+
)
|
| 31 |
+
MODELS[MODEL_NAME] = synthesizer
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
def tts(text: str, model_name: str):
|
| 35 |
+
print(text, model_name)
|
| 36 |
+
synthesizer = MODELS.get(model_name, None)
|
| 37 |
+
if synthesizer is None:
|
| 38 |
+
raise NameError("model not found")
|
| 39 |
+
wavs = synthesizer.tts(text)
|
| 40 |
+
# output = (synthesizer.output_sample_rate, np.array(wavs))
|
| 41 |
+
# return output
|
| 42 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
|
| 43 |
+
synthesizer.save_wav(wavs, fp)
|
| 44 |
+
return fp.name
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
iface = gr.Interface(
|
| 48 |
+
fn=tts,
|
| 49 |
+
inputs=[
|
| 50 |
+
gr.inputs.Textbox(
|
| 51 |
+
label="Input",
|
| 52 |
+
default="Привіт, як твої справи?",
|
| 53 |
+
),
|
| 54 |
+
gr.inputs.Radio(
|
| 55 |
+
label="Pick a TTS Model",
|
| 56 |
+
choices=MODEL_NAMES,
|
| 57 |
+
),
|
| 58 |
+
],
|
| 59 |
+
outputs=gr.outputs.Audio(label="Output"),
|
| 60 |
+
title="🐸💬 - Coqui TTS",
|
| 61 |
+
theme="huggingface",
|
| 62 |
+
description="🐸💬 - a deep learning toolkit for Text-to-Speech, battle-tested in research and production",
|
| 63 |
+
article="more info at https://github.com/coqui-ai/TTS",
|
| 64 |
+
)
|
| 65 |
+
iface.launch()
|
packages.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
libsndfile1-dev
|