Spaces:

sylus256
/

BeatDrop-Luganda-TTS

Configuration error

3.02 kB

Ctrl+K

1 contributor

Rename README.md to import gradio as gr import torchaudio from speechbrain.pretrained import Tacotron2, HIFIGAN tacotron = Tacotron2.from_hparams("Sunbird/tts-tacotron2-lug", savedir="/data/tts") vocoder = HIFIGAN.from_hparams("speechbrain/tts-hifigan-ljspeech", savedir="/data/voc") def speak(text): mel, _, _ = tacotron.encode_text(text) wav = vocoder.decode_batch(mel) torchaudio.save("out.wav", wav.squeeze(1).cpu(), 22050) return "out.wav" gr.Interface( fn=speak, inputs=gr.Textbox(value="Biitidulopu ye namba wani", label="Luganda"), outputs=gr.Audio(type="filepath"), title="BeatDrop Luganda TTS" ).launch()

a8c1a2e verified about 1 month ago

import gradio as gr import torchaudio from speechbrain.pretrained import Tacotron2, HIFIGAN tacotron = Tacotron2.from_hparams("Sunbird
Rename README.md to import gradio as gr import torchaudio from speechbrain.pretrained import Tacotron2, HIFIGAN tacotron = Tacotron2.from_hparams("Sunbird/tts-tacotron2-lug", savedir="/data/tts") vocoder = HIFIGAN.from_hparams("speechbrain/tts-hifigan-ljspeech", savedir="/data/voc") def speak(text): mel, _, _ = tacotron.encode_text(text) wav = vocoder.decode_batch(mel) torchaudio.save("out.wav", wav.squeeze(1).cpu(), 22050) return "out.wav" gr.Interface( fn=speak, inputs=gr.Textbox(value="Biitidulopu ye namba wani", label="Luganda"), outputs=gr.Audio(type="filepath"), title="BeatDrop Luganda TTS" ).launch() about 1 month ago
.gitattributes

1.52 kB
initial commit about 1 month ago
app.py

0 Bytes
import gradio as gr from transformers import VitsModel, AutoTokenizer import torch import numpy as np from scipy.io.wavfile import write as write_wav import tempfile # Load Luganda MMS-TTS model_id = "facebook/mms-tts-lug" tokenizer = AutoTokenizer.from_pretrained(model_id) model = VitsModel.from_pretrained(model_id) def tts_luganda(text): if not text.strip(): return None, "Nyiga ebigambo mu Luganda." inputs = tokenizer(text, return_tensors="pt") with torch.no_grad(): output = model(**inputs).waveform audio = output.squeeze().cpu().numpy() sr = model.config.sampling_rate # 16kHz audio = (audio * 32767).astype(np.int16) tmp = tempfile.NamedTemporaryFile(suffix=".wav", delete=False) write_wav(tmp.name, sr, audio) return tmp.name, f"Generated: {len(text)} chars" examples = [ "Oli otya? Ndi bulungi, webale.", "Sunbird AI ekola tekinologiya ku nnimi za Uganda.", "BeatDrop Luganda TTS ekuwa eddoboozi ery'obuntu.", ] demo = gr.Interface( fn=tts_luganda, inputs=gr.Textbox(label="Wandiika mu Luganda", placeholder="Wandiika wano...", lines=3), outputs=[gr.Audio(label="Eddoboozi"), gr.Textbox(label="Status")], title="BeatDrop - Luganda TTS", description="Demo ya text-to-speech mu Luganda ekozesebwa facebook/mms-tts-lug (Meta MMS).", examples=examples, allow_flagging="never" ) if __name__ == "__main__": demo.launch() about 1 month ago