Developer guide
What is the difference between a video transcription API and a social video analysis API?
A video transcription API returns spoken words, either from existing captions or speech recognition. A social video analysis API interprets a post’s videos and images, audio, captions, and text to answer a question about its actual content. It can identify what is shown or demonstrated, not just repeat words or surrounding post details.
Choose by the output
Transcription captures words; analysis interprets content.
Transcription is useful for searching or quoting speech. Content analysis is useful when you need to know what a post shows, explains, or demonstrates.
A transcript does not describe what is shown
A transcript captures spoken words, but not an unspoken demonstration or on-screen label. Content analysis can interpret those details from the post itself.
For speech
Use transcription to work with spoken words.
Transcription fits tasks such as finding videos that mention a phrase, quoting a speaker, or searching spoken content. It returns captions already available with a video or words recognized from audio.
For the whole post
Ask about what the post shows, says, or demonstrates.
In Mamitsu’s captured YouTube example, the prompt asks what a tutorial teaches. Mamitsu answers with five steps and the tools named in each. It identifies meaning in the video, rather than just repeating the speech.
Mamitsu interprets the post’s actual videos and images, audio, captions, and text to identify what is shown, explained, or demonstrated. Its answer is based on the content itself. See the API reference.
Check important details against the source
Analysis can only report details that can be understood from a post. When accuracy matters, check the answer against the original content.
FAQ
Questions developers ask about video content.
Does a video transcription API always use speech recognition?
No. Some services retrieve captions already supplied with the video; others use speech recognition to produce a transcript from audio.
How is content analysis different from post metadata?
Post metadata describes or identifies a post. Content analysis interprets what is shown, said, or written in the post to answer a specific question.
Can Mamitsu return structured information?
Yes. Mamitsu can organize its answer into structured fields suited to the question.
Build with Mamitsu