app.py

from transformers import pipeline
import gradio as gr

asr = pipeline(task="automatic-speech-recognition",
               model="openai/whisper-medium")

# Especificar el idioma de salida en español
asr.model.config.forced_decoder_ids = asr.tokenizer.get_decoder_prompt_ids(language="spanish", task="transcribe")

demo = gr.Blocks()

def transcribe_long_form(filepath):
    if filepath is None:
        gr.Warning("No audio found, please retry.")
        return ""
    output = asr(
      filepath,
      max_new_tokens=256,
      chunk_length_s=30,
      batch_size=8,
    )
    return output["text"]

ner = pipeline("ner",
               model="mrm8488/bert-spanish-cased-finetuned-ner",
)

def get_ner(input_text):
    if input_text is None:
        gr.Warning("No transcription found, please retry.")
        return {"text": "", "entities": ""}
    output = ner(input_text)
    return {"text": input_text, "entities": output}

def main(filepath):
    transcription = transcribe_long_form(filepath)
    ner = get_ner(transcription)
    return transcription, ner

mic_transcribe = gr.Interface(
    fn=main,
    inputs=gr.Audio(sources="microphone",
                    type="filepath"),
    outputs=[gr.Textbox(label="Transcription", lines=3),
             gr.HighlightedText(label="Text with entities")],
    title="Transcribir audio desde grabación",
    description="Transcripción de audio grabado desde micrófono.",
    allow_flagging="never")

file_transcribe = gr.Interface(
    fn=main,
    inputs=gr.Audio(sources="upload",
                    type="filepath"),
    outputs=[gr.Textbox(label="Transcription", lines=3),
             gr.HighlightedText(label="Text with entities")],
    title="Transcribir audio desde archivo",
    description="Transcripción a partir de un archivo de audio.",
    allow_flagging="never",
)

with demo:
    gr.TabbedInterface(
        [mic_transcribe,
         file_transcribe],
        ["Transcribe Microphone",
         "Transcribe Audio File"],
    )
demo.launch()