Hi,
When I generate TTS via the API, the output file size is about twice as large as when using the Playground. Is there a way to avoid having to compress the audio locally?
Can I set the bitrate or audio quality directly in the API request? If not, I’d really appreciate it if such an option could be added. Right now, TTS is taking up too much space on my disk.
Thanks in advance!
Welcome @descar
You can set the response_format to mp3, which will likely have the smallest file size of all formats for any given audio snippet.
Still too large, 1MB ~ 70 sek of dialog, where you have like 1000 users firebase will charge me a lot, I need to get half of this size. I am arleady try to use ffmpeg, but it is not supported starting this month. Is there a way to request such feature from Open AI?
IMO it’s better to store file on a s3 bucket than firebase. Store the path to file on the bucket on firebase instead.
Yes, it is ~15% cheaper but you need to make more configuration and deal with CORS which sometimes is getting me sick. Best option would be that open ai let you set bitrate in request, then I can get like 50% less cost from firebase, this is urgent for me.