yt_tool.py

import torch
import yt_dlp as youtube_dl
from transformers import pipeline
from transformers.pipelines.audio_utils import ffmpeg_read
from langchain_core.tools import tool

import tempfile
import os

# credit https://huggingface.co/spaces/hf-audio/whisper-large-v3

MODEL_NAME = "openai/whisper-tiny.en"
BATCH_SIZE = 8
FILE_LIMIT_MB = 1000
YT_LENGTH_LIMIT_S = 3600  # limit to 1 hour YouTube files

device = "mps" if torch.mps.is_available() else "cpu"

speech_recognition_pipe = pipeline(
    task="automatic-speech-recognition",
    model=MODEL_NAME,
    chunk_length_s=30,
    device=device,
)

def download_yt_audio(yt_url, filename):
    info_loader = youtube_dl.YoutubeDL()

    try:
        info = info_loader.extract_info(yt_url, download=False)
    except youtube_dl.utils.DownloadError as err:
        raise str(err)

    file_length = info["duration_string"]
    file_h_m_s = file_length.split(":")
    file_h_m_s = [int(sub_length) for sub_length in file_h_m_s]

    if len(file_h_m_s) == 1:
        file_h_m_s.insert(0, 0)
    if len(file_h_m_s) == 2:
        file_h_m_s.insert(0, 0)
    file_length_s = file_h_m_s[0] * 3600 + file_h_m_s[1] * 60 + file_h_m_s[2]

    if file_length_s > YT_LENGTH_LIMIT_S:
        yt_length_limit_hms = time.strftime("%HH:%MM:%SS", time.gmtime(YT_LENGTH_LIMIT_S))
        file_length_hms = time.strftime("%HH:%MM:%SS", time.gmtime(file_length_s))
        raise f"Maximum YouTube length is {yt_length_limit_hms}, got {file_length_hms} YouTube video."

    ydl_opts = {"outtmpl": filename, "format": "worstvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best"}

    with youtube_dl.YoutubeDL(ydl_opts) as ydl:
        try:
            ydl.download([yt_url])
        except youtube_dl.utils.ExtractorError as err:
            raise str(err)

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

@tool
def yt_transcribe(yt_url, max_filesize=75.0):
    """
        Transcribes the audio from a given YouTube video URL.

        Args:
            yt_url (str): The URL of the YouTube video.
            max_filesize (float, optional): The maximum allowed filesize of the video in MB.
                                             Defaults to 75.0.

        Returns:
            tuple: A tuple containing:
                   - str: An HTML embed string for the YouTube video.
                   - str: The transcribed text of the video's audio.
        """
    html_embed_str = _return_yt_html_embed(yt_url)

    with tempfile.TemporaryDirectory() as tmpdirname:
        filepath = os.path.join(tmpdirname, "video.mp4")
        download_yt_audio(yt_url, filepath)
        with open(filepath, "rb") as f:
            inputs = f.read()

    inputs = ffmpeg_read(inputs, speech_recognition_pipe.feature_extractor.sampling_rate)
    inputs = {"array": inputs, "sampling_rate": speech_recognition_pipe.feature_extractor.sampling_rate}

    text = speech_recognition_pipe(inputs, batch_size=8, return_timestamps=True)["text"]

    return html_embed_str, text