spotify_transcript.py

import requests
import json

CLIENT_ID = "7fb1be85577a44a99036ae5afeded772"
CLIENT_SECRET = "fc05d3027d754b149c8d8d82aca9daee"
EPISODE_ID = "6FfOT7TDLNzdD1M81SChga"


def get_access_token():
    url = "https://accounts.spotify.com/api/token"
    resp = requests.post(url, data={
        "grant_type": "client_credentials",
        "client_id": CLIENT_ID,
        "client_secret": CLIENT_SECRET,
    })
    resp.raise_for_status()
    return resp.json()["access_token"]


def get_episode_info(token, episode_id):
    url = f"https://api.spotify.com/v1/episodes/{episode_id}"
    resp = requests.get(url, headers={"Authorization": f"Bearer {token}"})
    resp.raise_for_status()
    return resp.json()


def main():
    print("Authenticating with Spotify...")
    token = get_access_token()

    print("Fetching episode info...")
    episode = get_episode_info(token, EPISODE_ID)

    print(f"\nPodcast: {episode['show']['name']}")
    print(f"Episode: {episode['name']}")
    print(f"Duration: {episode['duration_ms'] // 60000} min")
    print(f"Description: {episode['description'][:300]}")
    print(f"Release: {episode['release_date']}")
    print(f"Audio Preview URL: {episode.get('audio_preview_url', 'N/A')}")

    # Try undocumented transcript endpoint
    print("\nAttempting to fetch transcript...")
    transcript_url = f"https://spclient.wg.spotify.com/transcript-read-along/v2/episode/{EPISODE_ID}"
    resp = requests.get(transcript_url, headers={"Authorization": f"Bearer {token}"})
    if resp.status_code == 200:
        data = resp.json()
        print("Transcript found!\n")
        if "section" in data:
            for section in data["section"]:
                for line in section.get("startMs", []):
                    print(line.get("words", ""))
        else:
            print(json.dumps(data, indent=2, ensure_ascii=False))
    else:
        print(f"Transcript not available via this method (status {resp.status_code})")
        print("\nSpotify does not expose transcripts through the public Web API.")
        print("Alternatives:")
        print("  1. If the podcast is on YouTube, we can grab the transcript there")
        print("  2. Download the audio and use Whisper for speech-to-text")
        print("  3. Check the podcast's website for published transcripts")


if __name__ == "__main__":
    main()