search

MacPaw / OpenAI

Swift community driven package for OpenAI public API
MIT License
2.11k stars 353 forks source link

fixed verboseJson encoding #236

Open Frank-Buss opened 2 weeks ago

Frank-Buss commented 2 weeks ago

What

This fixes the verboseJson encoding. I tested it with verboseJson, and works still with text as well.

Why

I got this error: Transcription failed: APIErrorResponse(error: OpenAI.APIError(message: "[{\'type\': \'enum\', \'loc\': (\'body\', \'response_format\'), \'msg\': \"Input should be \'json\', \'text\', \'vtt\', \'srt\' or \'verbose_json\'\", \'input\': \'verboseJson\', \'ctx\': {\'expected\': \"\'json\', \'text\', \'vtt\', \'srt\' or \'verbose_json\'\"}}]", type: "invalid_request_error", param: nil, code: nil))

because verboseJson was not encoded as verbose_json.

Affected Areas

AudioTranscriptionQuery

sonarcloud[bot] commented 2 weeks ago

Quality Gate Passed Quality Gate passed

Issues
3 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

Frank-Buss commented 1 week ago

Anyone who wants to review it? I need this for my project. I can use my branch, but would be better to integrate this change, so it works for everybody. BTW, I needed it because Whisper returns nonsense for silence, which is easily fixed by testing no_speech_prob for > 0.4 with the verbose json response.