Limitations of Google Speech API

Question

Can someone share the limitation of google Speech API ?

We heard below limitations so far:

1. Can’t transcribe audio files more than 60 seconds
2. Can’t handle dual channel (agent and customer)

Thanks

Answer 1

There are limits besides the amount of time for a single request. I finally hit a limit recently after a lot of experimentation; as soon as a request starts, it's cut short. On a desktop web browser (Chrome), there's a limit for each user (Primiano Tucci works at Google, so I'd trust him!), and you can't buy more time with the API.

However, as you could see from the first link, the exception appears to be Chrome on Android, which apparently has no limitations because it accesses an internal API for speech recognition when visiting web pages that use the Web Speech API.

For desktop browsers, you're out of luck unless you can assume users don't mind a limit. Even Annyang and voice dictation in Google Docs aren't working for me at the moment, while online voice recognition services that don't go through Google are working.