Video Transcription with Google Colab

Navigate to Google Colab and create a new notebook.

Copy the following code and paste it into the Colab notebook cell:

!pip install SpeechRecognition
from google.colab import files
import subprocess
import os
import speech_recognition as sr
import datetime

def extract_audio():
    print("Upload a video file...")
    uploaded = files.upload()
    video_filename = list(uploaded.keys())[0]
    audio_filename = os.path.splitext(video_filename)[0] + ".wav"
    subprocess.run(["ffmpeg", "-i", video_filename, "-ac", "1", "-ar", "16000", audio_filename], check=True)
    print("Extraction completed! Download the audio:")
    files.download(audio_filename)

def transcribe_audio():
    print("Upload an audio file for transcription...")
    uploaded = files.upload()
    audio_filename = list(uploaded.keys())[0]
    subtitle_filename = os.path.splitext(audio_filename)[0] + ".srt"
    recognizer = sr.Recognizer()
    with sr.AudioFile(audio_filename) as source:
        duration = int(source.DURATION) # Total duration of the audio in seconds
        chunk_size = 5 # Seconds for each segment
        subtitles = []
        for i in range(0, duration, chunk_size):
            source_audio = recognizer.record(source, duration=chunk_size)
            try:
                text = recognizer.recognize_google(source_audio, language="it-IT")

                start_time = str(datetime.timedelta(seconds=i)) + ",000"
                end_time = str(datetime.timedelta(seconds=i + chunk_size)) + ",000"
                subtitles.append(f"{len(subtitles) + 1}\n{start_time} --> {end_time}\n{text}\n\n")
            except sr.UnknownValueError:
                continue
            except sr.RequestError:
                print("Error in transcription request.")
                return
    with open(subtitle_filename, "w") as f:
        f.writelines(subtitles)
    print("Transcription completed! Download the subtitle file:")
    files.download(subtitle_filename)

print("Choose an operation:")
print("1: Extract audio from a video")
print("2: Transcribe an audio file into subtitles")
choice = input("Enter 1 or 2: ")

if choice == "1":
    extract_audio()
elif choice == "2":
    transcribe_audio()
else:
    print("Invalid choice. Try again.")

Click on the play button (run cell) or press Shift+Enter to execute the code.

When prompted, enter the number of the function you want to run:

• Option 1: Extract audio from a video file
• Option 2: Transcribe an audio file into subtitles

Follow the prompts to upload your video or audio file. The script will:

• For Option 1: Extract audio and provide a download link for the .wav file
• For Option 2: Create a .srt subtitle file and provide a download link

To change the recognition language, modify the language parameter in the line:

text = recognizer.recognize_google(source_audio, language="it-IT")

For example, use:

• language="en-US" for US English
• language="es-ES" for Spanish
• language="fr-FR" for French
• language="de-DE" for German

To adjust the length of each subtitle segment, change the chunk_size variable:

chunk_size = 5 # Seconds for each segment

A smaller value will create more, shorter subtitles; a larger value will create fewer, longer subtitles.