app.py

import gradio as gr
import asr
import tts
import util

mms_transcribe = gr.Interface(
    fn=asr.transcribe,
    inputs=[
        gr.Audio(),
        gr.Dropdown(
            choices=[model for model in asr.models_info],
            label="Select a Model for ASR",
            value="ixxan/wav2vec2-large-mms-1b-uyghur-latin",
            interactive=True
        ),
    ],
    outputs=[
        gr.Textbox(label="Uyghur Arabic Transcription"),
        gr.Textbox(label="Uyghur Latin Transcription"),
    ],
    #examples=util.asr_examples,
    title="Speech-to-text",
    description=(
        "Transcribe Uyghur speech audio from a microphone or input file."
    ),
    allow_flagging="never",
)

mms_synthesize = gr.Interface(
    fn=tts.synthesize,
    inputs=[
        gr.Text(label="Input text"),
        gr.Dropdown(
            choices=[model for model in tts.models_info],
            label="Select a Model for TTS",
            value="Meta-MMS",
            interactive=True
        )
    ],
    outputs=[
        gr.Audio(label="Generated Audio"),
    ],
    #examples=util.tts_examples,
    title="Text-to-speech",
    description=(
        "Generate audio from input Uyghur text."
        ),
    allow_flagging="never",
)

tabbed_interface = gr.TabbedInterface(
    [mms_transcribe, mms_synthesize],
    ["Speech-to-text", "Text-to-speech"],
)

with gr.Blocks() as demo:
    gr.Markdown("Comparision of STT and TTS models for Uyghur language.")
    tabbed_interface.render()

if __name__ == "__main__":
    demo.queue()
    demo.launch()