app.py

# from speechbrain.inference.ASR import EncoderASR
# import gradio as gr

# model = EncoderASR.from_hparams("speechbrain/asr-wav2vec2-dvoice-wolof")

# def transcribe(audio):
#   return model.transcribe_file(audio.name)


# demo = gr.Interface(fn=transcribe, inputs="file", outputs="text", 
#                    title="Transcription automatique du wolof", 
#              description="Ce modèle transcrit un fichier audio en wolof en texte en utilisant l'alphabet latin.",              
#              input_label="Audio en wolof", 
#              output_label="Transcription alphabet latin"
#                    )

# demo.launch()


from speechbrain.inference.ASR import EncoderASR
import gradio as gr

model = EncoderASR.from_hparams("speechbrain/asr-wav2vec2-dvoice-wolof")

def transcribe(audio):
    return model.transcribe_file(audio.name)

def transcribe_microphone(audio):
    return model.transcribe_array(audio)

demo = gr.Interface(
    fn=[transcribe, transcribe_microphone], 
    inputs=["file", gr.inputs.Microphone(audio=False)], 
    outputs="text", 
    title="Transcription automatique du wolof", 
    description="Ce modèle transcrit un fichier audio en wolof en texte en utilisant l'alphabet latin.",
    input_labels=["Audio en wolof", "Microphone (parlez en wolof)"], 
    output_label="Transcription alphabet latin"
)

demo.launch()