app.py

import torch
import os

import gradio as gr
import pytube as pt
from speechbox import ASRDiarizationPipeline

MODEL_NAME = "openai/whisper-small"

device = 0 if torch.cuda.is_available() else "cpu"
HF_TOKEN = os.environ.get("HF_TOKEN")

pipe = ASRDiarizationPipeline.from_pretrained(
    asr_model=MODEL_NAME,
    device=device,
    use_auth_token=HF_TOKEN,
)

def tuple_to_string(start_end_tuple, ndigits=1):
    return str((round(start_end_tuple[0], ndigits), round(start_end_tuple[1], ndigits)))


def format_as_transcription(raw_segments, with_timestamps=False):
    if with_timestamps:
        return "\n\n".join([chunk["speaker"] + " " + tuple_to_string(chunk["timestamp"]) +  chunk["text"] for chunk in raw_segments])
    else:
        return "\n\n".join([chunk["speaker"] + chunk["text"] for chunk in raw_segments])


def transcribe(file_upload, with_timestamps):
    raw_segments = pipe(file_upload)
    transcription = format_as_transcription(raw_segments, with_timestamps=with_timestamps)
    return transcription


def _return_yt_html_embed(yt_url):
    video_id = yt_url.split("?v=")[-1]
    HTML_str = (
        f'<center> <iframe width="500" height="320" src="https://www.youtube.com/embed/{video_id}"> </iframe>'
        " </center>"
    )
    return HTML_str


def yt_transcribe(yt_url, with_timestamps):
    yt = pt.YouTube(yt_url)
    html_embed_str = _return_yt_html_embed(yt_url)
    stream = yt.streams.filter(only_audio=True)[0]
    stream.download(filename="audio.mp3")

    text = pipe("audio.mp3")

    return html_embed_str, format_as_transcription(text, with_timestamps=with_timestamps)


demo = gr.Blocks()

mf_transcribe = gr.Interface(
    fn=transcribe,
    inputs=[
        gr.inputs.Audio(source="upload", type="filepath"),
        gr.Checkbox(label="With timestamps?", value=True),
    ],
    outputs="text",
    layout="horizontal",
    theme="huggingface",
    title="Whisper Speaker Diarization: Transcribe Audio",
    description=(
        "Transcribe audio files with speaker diarization using [🤗 Speechbox](https://github.com/huggingface/speechbox/). "
        "Demo uses the pre-trained checkpoint [Whisper Small](https://huggingface.co/openai/whisper-small) for the ASR "
        "transcriptions and [pyannote.audio](https://huggingface.co/pyannote/speaker-diarization) to label the speakers."
        "\n\n"
        "Check out the repo here: https://github.com/huggingface/speechbox/"
    ),
    examples=[
        ["./processed.wav", True],
        ["./processed.wav", False],
    ],
    allow_flagging="never",
)

yt_transcribe = gr.Interface(
    fn=yt_transcribe,
    inputs=[
        gr.inputs.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL"),
        gr.Checkbox(label="With timestamps?", value=True),
    ],
    outputs=["html", "text"],
    layout="horizontal",
    theme="huggingface",
    title="Whisper Speaker Diarization: Transcribe YouTube",
    description=(
        "Transcribe YouTube videos with speaker diarization using [🤗 Speechbox](https://github.com/huggingface/speechbox/). "
        "Demo uses the pre-trained checkpoint [Whisper Tiny](https://huggingface.co/openai/whisper-tiny) for the ASR "
        "transcriptions and [pyannote.audio](https://huggingface.co/pyannote/speaker-diarization) to label the speakers."
        "\n\n"
        "Check out the repo here: https://github.com/huggingface/speechbox/"
    ),
    examples=[
        ["https://www.youtube.com/watch?v=9dAWIPixYxc", True],
        ["https://www.youtube.com/watch?v=9dAWIPixYxc", False],
    ],
    allow_flagging="never",
)

with demo:
    gr.TabbedInterface([mf_transcribe, yt_transcribe], ["Transcribe Audio", "Transcribe YouTube"])

demo.launch(enable_queue=True)