Whisper Transcription

OpenAI Whisper provides accurate speech-to-text with word-level timestamps.

Installation

pip install openai-whisper

Model Selection

Use the tiny model for fast transcription - it's sufficient for most tasks and runs much faster:

Model	Size	Speed	Accuracy
tiny	39 MB	Fastest	Good for clear speech
base	74 MB	Fast	Better accuracy
small	244 MB	Medium	High accuracy

Recommendation: Start with tiny - it handles clear interview/podcast audio well.

Basic Usage with Word Timestamps

import whisper
import json

def transcribe_with_timestamps(audio_path, output_path):
    """
    Transcribe audio and get word-level timestamps.

    Args:
        audio_path: Path to audio/video file
        output_path: Path to save JSON output
    """
    # Use tiny model for speed
    model = whisper.load_model("tiny")

    # Transcribe with word timestamps
    result = model.transcribe(
        audio_path,
        word_timestamps=True,
        language="en"  # Specify language for better accuracy
    )

    # Extract words with timestamps
    words = []
    for segment in result["segments"]:
        if "words" in segment:
            for word_info in segment["words"]:
                words.append({
                    "word": word_info["word"].strip(),
                    "start": word_info["start"],
                    "end": word_info["end"]
                })

    with open(output_path, "w") as f:
        json.dump(words, f, indent=2)

    return words

Detecting Specific Words

def find_words(transcription, target_words):
    """
    Find specific words in transcription with their timestamps.

    Args:
        transcription: List of word dicts with 'word', 'start', 'end'
        target_words: Set of words to find (lowercase)