|
|
import gradio as gr |
|
|
from transformers import VitsModel, AutoTokenizer |
|
|
import torch |
|
|
import logging |
|
|
import spaces |
|
|
from typing import Tuple, Optional |
|
|
import numpy as np |
|
|
|
|
|
logging.basicConfig(level=logging.INFO) |
|
|
logger = logging.getLogger(__name__) |
|
|
|
|
|
if torch.cuda.is_available(): |
|
|
device = "cuda" |
|
|
logger.info("Using CUDA for inference.") |
|
|
elif torch.backends.mps.is_available(): |
|
|
device = "mps" |
|
|
logger.info("Using MPS for inference.") |
|
|
else: |
|
|
device = "cpu" |
|
|
logger.info("Using CPU for inference.") |
|
|
|
|
|
languages = ["bambara", "boomu", "dogon", "pular", "songhoy", "tamasheq"] |
|
|
|
|
|
examples = { |
|
|
|
|
|
"bambara": "An filɛ ni ye yɔrɔ minna ni an ye an sigi ka a layɛ yala an bɛ ka baara min kɛ ɛsike a kɛlen don ka Ɲɛ wa ?", |
|
|
|
|
|
"boomu": "Vunurobe wozomɛ pɛɛ, Poli we zo woro han Deeɓenu wara li Deeɓenu faralo zuun. Lo we baba a lo wara yi see ɓa Zuwifera ma ɓa Gɛrɛkela wa.", |
|
|
|
|
|
"dogon": "Pɔɔlɔ, kubɔ lugo joo le, bana dɛin dɛin le, inɛw Ama titiyaanw le digɛu, Ama, emɛ babe bɛrɛ sɔɔ sɔi.", |
|
|
|
|
|
"pular": "Miɗo ndaarde saabe Laamɗo e saabe Iisaa Almasiihu caroyoowo wuurɓe e maayɓe oo, miɗo ndaardire saabe gartol makko ka num e Laamu makko", |
|
|
|
|
|
"songhoy": "Haya ka se beenediyo kokoyteraydi go hima nda huukoy foo ka fatta ja subaahi ka taasi goykoyyo ngu rezẽ faridi se", |
|
|
|
|
|
"tamasheq": "Toḍă tăfukt ɣas, issăɣră-dd măssi-s n-ašĕkrĕš ănaẓraf-net, inn'-as: 'Ǝɣĕr-dd inaxdimăn, tĕẓlĕd-asăn, sănt s-wi dd-ĕšrăynen har tĕkkĕd wi dd-ăzzarnen." |
|
|
|
|
|
} |
|
|
|
|
|
class MalianTTS: |
|
|
def __init__(self, model_name: str = "MALIBA-AI/malian-tts"): |
|
|
self.model_name = model_name |
|
|
self.models = {} |
|
|
self.tokenizers = {} |
|
|
self._load_models() |
|
|
|
|
|
def _load_models(self): |
|
|
"""Load all language models and tokenizers""" |
|
|
try: |
|
|
for lang in languages: |
|
|
logger.info(f"Loading model and tokenizer for {lang}...") |
|
|
self.models[lang] = VitsModel.from_pretrained( |
|
|
self.model_name, |
|
|
subfolder=f"models/{lang}" |
|
|
).to(device) |
|
|
self.tokenizers[lang] = AutoTokenizer.from_pretrained( |
|
|
self.model_name, |
|
|
subfolder=f"models/{lang}" |
|
|
) |
|
|
logger.info(f"Successfully loaded {lang}") |
|
|
except Exception as e: |
|
|
logger.error(f"Failed to load models: {str(e)}") |
|
|
raise Exception(f"Model loading failed: {str(e)}") |
|
|
|
|
|
def synthesize(self, language: str, text: str) -> Tuple[Optional[Tuple[int, np.ndarray]], Optional[str]]: |
|
|
"""Generate audio from text for the specified language""" |
|
|
if not text.strip(): |
|
|
return None, "Please enter some text to synthesize." |
|
|
|
|
|
try: |
|
|
model = self.models[language] |
|
|
tokenizer = self.tokenizers[language] |
|
|
|
|
|
inputs = tokenizer(text, return_tensors="pt").to(device) |
|
|
|
|
|
with torch.no_grad(): |
|
|
output = model(**inputs).waveform |
|
|
|
|
|
waveform = output.squeeze().cpu().numpy() |
|
|
sample_rate = model.config.sampling_rate |
|
|
|
|
|
return (sample_rate, waveform), None |
|
|
|
|
|
except Exception as e: |
|
|
logger.error(f"Error during inference for {language}: {str(e)}") |
|
|
return None, f"Error generating audio: {str(e)}" |
|
|
|
|
|
|
|
|
tts_system = MalianTTS() |
|
|
|
|
|
@spaces.GPU() |
|
|
def generate_audio(language: str, text: str) -> Tuple[Optional[Tuple[int, np.ndarray]], str]: |
|
|
""" |
|
|
Generate audio from text using the specified language model. |
|
|
""" |
|
|
if not text.strip(): |
|
|
return None, "Please enter some text to synthesize." |
|
|
|
|
|
try: |
|
|
audio_output, error_msg = tts_system.synthesize(language, text) |
|
|
if error_msg: |
|
|
logger.error(f"TTS generation failed: {error_msg}") |
|
|
return None, error_msg |
|
|
|
|
|
logger.info(f"Successfully generated audio for {language}") |
|
|
return audio_output, "Audio generated successfully!" |
|
|
|
|
|
except Exception as e: |
|
|
logger.error(f"Audio generation failed: {e}") |
|
|
return None, f"Error: {str(e)}" |
|
|
|
|
|
def load_example(language: str) -> str: |
|
|
"""Load example text for the selected language""" |
|
|
return examples.get(language, "No example available") |
|
|
|
|
|
def build_interface(): |
|
|
""" |
|
|
Builds the Gradio interface for Malian TTS. |
|
|
""" |
|
|
with gr.Blocks(title="MalianVoices") as demo: |
|
|
gr.Markdown( |
|
|
""" |
|
|
# MalianVoices: 🇲🇱 Text-to-Speech in Six Malian Languages |
|
|
|
|
|
Lightweight TTS for six Malian languages: **Bambara, Boomu, Dogon, Pular, Songhoy, Tamasheq**. |
|
|
|
|
|
- ✅ Real-time TTS with fast response |
|
|
|
|
|
## How to Use |
|
|
1. Pick a language from the dropdown |
|
|
2. Enter your text or load an example |
|
|
3. Click **"Generate Audio"** to listen |
|
|
""" |
|
|
) |
|
|
|
|
|
with gr.Row(): |
|
|
language = gr.Dropdown( |
|
|
choices=languages, |
|
|
label="Language", |
|
|
value="bambara" |
|
|
) |
|
|
|
|
|
with gr.Column(): |
|
|
text = gr.Textbox( |
|
|
label="Input Text", |
|
|
lines=5, |
|
|
placeholder="Type your text here..." |
|
|
) |
|
|
|
|
|
with gr.Row(): |
|
|
example_btn = gr.Button("Load Example") |
|
|
generate_btn = gr.Button("Generate Audio", variant="primary") |
|
|
|
|
|
audio_output = gr.Audio(label="Generated Audio", type="numpy") |
|
|
status_msg = gr.Textbox(label="Status", interactive=False) |
|
|
|
|
|
|
|
|
gr.Markdown( |
|
|
""" |
|
|
By [sudoping01](https://huggingface.co/sudoping01), from [sudoping01/malian-tts](https://huggingface.co/sudoping01/malian-tts). |
|
|
Fine-tuned on Meta's MMS, CC BY-NC 4.0, non-commercial. |
|
|
""" |
|
|
) |
|
|
|
|
|
|
|
|
generate_btn.click( |
|
|
fn=generate_audio, |
|
|
inputs=[language, text], |
|
|
outputs=[audio_output, status_msg] |
|
|
) |
|
|
|
|
|
example_btn.click( |
|
|
fn=load_example, |
|
|
inputs=language, |
|
|
outputs=text |
|
|
) |
|
|
|
|
|
return demo |
|
|
|
|
|
if __name__ == "__main__": |
|
|
logger.info("Starting the Gradio interface for MalianVoices TTS.") |
|
|
interface = build_interface() |
|
|
interface.launch() |
|
|
logger.info("Gradio interface running.") |
