Support tts model selection (tts-1, tts-1-1106, tts-1-hd, tts-1-hd-1106)

[mcint]

Support tts model selection to better make use of request request quotas. Support among tts-1, tts-1-1106, tts-1-hd, tts-1-hd-1106.

For at least me, the -1106 versions support 3k Req/min, versus 3 Req/min for the default.

[tomviner]

This would also allow testing out the canary version of tts (canary-tts), which was added a few days ago.

As tracked by my OpenAI models API scraper at https://github.com/tomviner/scrape-ai/blame/main/openai/api/models.txt#L14

[simonw]

Fixed in #5

[simonw]

This works:

ospeak -m tts-1-hd 'hello' 

This shows an error:

ospeak -m tts-1-hd2 'hello'

Error: The model `tts-1-hd2` does not exist