| import os | |
| os.system("pip install gradio==3.3") | |
| import gradio as gr | |
| import numpy as np | |
| import streamlit as st | |
| title = "Fairseq S2S" | |
| description = "Gradio Demo for fairseq S2S: speech-to-speech translation models. To use it, simply add your audio, or click one of the examples to load them. Read more at the links below." | |
| article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2107.05604' target='_blank'>Direct speech-to-speech translation with discrete units</a> | <a href='https://github.com/facebookresearch/fairseq/tree/main/examples/speech_to_speech' target='_blank'>Github Repo</a></p>" | |
| examples = [ | |
| ["enhanced_direct_s2st_units_audios_es-en_set2_source_12478_cv.flac","xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022"], | |
| ] | |
| io1 = gr.Interface.load("huggingface/facebook/xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022", api_key=st.secrets["api_key"]) | |
| def inference(text,model): | |
| outtext = io1(text) | |
| return outtext | |
| gr.Interface( | |
| inference, | |
| inputs=[ | |
| gr.Audio(source="microphone", type="filepath", optional=True, label="Input"), | |
| gr.Audio(source="upload", type="filepath", optional=True, label="Input"), | |
| gr.inputs.Dropdown(choices=["xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022"], default="xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022", type="value", label="Model"), | |
| ], | |
| gr.outputs.Audio(label="Output"), | |
| article=article, | |
| title=title, | |
| examples=examples, | |
| description=description).queue().launch() |