How is youtube transcription feature reliable
Hi, i want to use youtube transcription functionality on my website to provide this functionality to end users.i have some questions though, could you please provide me some information regarding:
1. I need to use
// Get transcript from any supported platform (YouTube, TikTok, Twitter) or file
const transcriptResult = await supadata.transcript({
or
// Get YouTube transcript
const transcript: Transcript = await supadata.youtube.transcript({
2. What is the percentage of youtube videos that will return me transcriptions? (in english)
3. What happens if there is no transcriptions for the video?
4. the pricing will be always 1 transcript = 1 credit ? Or also a 1 generated transcript minute = 2 credits is possible? for youtube videos?
5. If queue is started to get video transcriptions - how long it takes?
So basically - what is the approach you can suggest me to use your api, to be able to calculate the token usage and be sure the customer will get it’s transcription in reasonable time?
Thank you.
Log in to comment and vote
Comments1
Rafal
Jul 16, 2025
Hi Konstantin, thanks for asking! Here are the answers:
1. The
supadata.transcript()is recommended oversupadata.youtube.transcript()as the new version with more functionality. The first one will be eventually deprecated.2. Based on the API results we estimate that “worthwhile videos” in the categories of science, entertainment, news (by “worthwhile” I mean non-music, videos with more than a dozen views, not in rare languages etc) the percentage of videos without transcripts is around 10-15%.
3. If there is no transcript and you chose to not generate one (
mode = native) then the API will return code HTTP 206. By default,supadata.transcript()will generate a transcript if there is none.4. For
supadata.youtube.transcript()pricing is always 1 transcript = 1 credit. Forsupadata.transcript()pricing depends if the transcript was generated or not.5. The AI transcription currently takes approximately 1s per minute of video. This number is subject to change as we’re continuously working on improving performance since the release two weeks ago.
Hope that helps! Feel free to also contact support@supadata.ai if you have questions specific to your use case.