Spaces:
Runtime error
Runtime error
| import gradio as gr | |
| import librosa | |
| from optimum.onnxruntime import ORTModelForSeq2SeqLM | |
| from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC | |
| from transformers import AutoTokenizer, AutoModelForSeq2SeqLM | |
| import torch | |
| # load model and processor | |
| processor = Wav2Vec2Processor.from_pretrained("jonatasgrosman/wav2vec2-large-xlsr-53-english") | |
| model = Wav2Vec2ForCTC.from_pretrained("jonatasgrosman/wav2vec2-large-xlsr-53-english") | |
| tokenizer = AutoTokenizer.from_pretrained("icon-it-tdtu/mt-en-vi-optimum") | |
| model_lm = ORTModelForSeq2SeqLM.from_pretrained("icon-it-tdtu/mt-en-vi-optimum") | |
| def process_audio_file(file): | |
| data, sr = librosa.load(file) | |
| if sr != 16000: | |
| data = librosa.resample(data, sr, 16000) | |
| print(data.shape) | |
| inputs = processor(data, sampling_rate=16000, return_tensors="pt", padding=True) | |
| return inputs | |
| def interpret(micro, file): | |
| if file is not None and micro is None: | |
| input_audio = file | |
| elif file is None and micro is not None: | |
| input_audio = micro | |
| else: | |
| input_audio = file | |
| inputs = process_audio_file(input_audio) | |
| with torch.no_grad(): | |
| output_logit = model(inputs.input_values).logits | |
| pred_ids = torch.argmax(output_logit, dim=-1) | |
| text = processor.batch_decode(pred_ids)[0].lower() | |
| print(text) | |
| translate_text = translate(text) | |
| return translate_text | |
| def translate(text): | |
| batch = tokenizer([text], return_tensors="pt") | |
| generated_ids = model_lm.generate(**batch) | |
| translated_text = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0] | |
| return translated_text | |
| # Set the starting state to an empty string | |
| iface = gr.Interface( | |
| fn=interpret, | |
| title="Interpret English to Vietnamese", | |
| description="A simple interface to interpret from spoken English to Vietnamese.", | |
| article="Author: <a href=\"https://huggingface.co/vumichien\">Vu Minh Chien</a>.", | |
| inputs=[ | |
| gr.Audio(source="microphone", type="filepath", streaming=False, label="Microphone"), | |
| gr.Audio(source="upload", type="filepath", label="File"), | |
| ], | |
| outputs=gr.Textbox(label="Text output") | |
| , | |
| examples=[[None, "samples/sample-000000.mp3"], [None, "samples/sample-000001.mp3"]], | |
| ) | |
| iface.launch(debug=True) |