**I am facing an TTS-audio conundrum.**

I am facing an TTS-audio conundrum.
The TTS voices offered by openAI all sound much cleaner and more voluminous than those via the API. The same voice over the API sounds tinny, as if there is some kind of distortion filter over it.
Does anyone have any idea why this might be?
Many thanks from berlin
Jan

Are you experiencing this issue when you set the quality to tts-hd as well, or are you using regular tts?

Hi! Yes, the settings are set to HD.
But thanks in advance for thinking along. :wink:
Best, Jan

1 Like

I wanted to send you an audio sample via Dropbox. But links don’t seem to be allowed here. :frowning:

You need to participate a bit before links as we get a lot of spammers.

I remember someone else mentioning quality. Is the new Voice Engine going to be rolled into TTS I wonder?

The German sample voices on the OpenAI site sound good. But unfortunately not really via API.
IMG_8691
or we are doing something wrong…???.. who knows :wink:

Welcome @dommel1

The response_format param for createSpeech request defaults to mp3 which is the worst of all available options for audio quality.

Setting it to aac should slightly improve the quality but it’s still lossy.

If you want to go for the best audio quality use opus , flac , wav , or pcm.

yes, thank you for thinking about it! we tried the different audio formats without any significant improvement. But the “metallic soundeffect” unfortunately remained. Greetings from Berlin Jan