asr-inference / app.py
ssolito's picture
Pause GPU
ab6d2b5 verified
import gradio as gr
from whisper import generate
from AinaTheme import theme
#import spaces
USE_V5 = False
#@spaces.GPU
def transcribe(inputs, model_version):
if inputs is None:
raise gr.Error("Cap fitxer d'脿udio introduit! Si us plau pengeu un fitxer "\
"o enregistreu un 脿udio abans d'enviar la vostra sol路licitud")
use_v5 = model_version == "v0.5"
return generate(audio_path=inputs, use_v5=use_v5)
description_string = "Transcripci贸 autom脿tica de micr貌fon o de fitxers d'脿udio.\n Aquest demostrador s'ha desenvolupat per"\
" comprovar els models de reconeixement de parla per a m贸bils."
def clear():
return None, "v1.0"
with gr.Blocks() as demo:
gr.Markdown(description_string)
with gr.Row():
with gr.Column(scale=1):
model_version = gr.Dropdown(label="Model Version", choices=["v1.0", "v0.5"], value="v1.0")
input = gr.Audio(sources=["upload", "microphone"], type="filepath", label="Audio")
with gr.Column(scale=1):
output = gr.Textbox(label="Output", lines=8)
with gr.Row(variant="panel"):
clear_btn = gr.Button("Clear")
submit_btn = gr.Button("Submit", variant="primary")
submit_btn.click(fn=transcribe, inputs=[input, model_version], outputs=[output])
clear_btn.click(fn=clear, inputs=[], outputs=[input, model_version], queue=False)
if __name__ == "__main__":
demo.launch()