fix(voice): send language and prompt in streamed STT

[sylvesterkaczmarek]

Summary

Honor STTModelSettings.language and STTModelSettings.prompt in streamed OpenAI transcription sessions.

The non-streaming STT path already forwards both settings to audio.transcriptions.create(), but the streamed path previously sent only the transcription model. Callers could therefore configure language or prompt and still have those values silently omitted from the Realtime transcription session.

Fix

• include prompt when explicitly configured
• use plural languages: [value] for gpt-live-transcribe
• retain singular language for gpt-transcribe and existing streamed transcription models such as gpt-4o-transcribe
• omit language/prompt keys when unset, preserving the existing default session payload
• leave temperature unchanged because it is not part of the current Realtime transcription-session configuration

Test plan

Focused coverage verifies that:

• gpt-4o-transcribe receives singular language and configured prompt
• gpt-transcribe receives singular language and configured prompt
• gpt-live-transcribe receives plural languages: [value]
• unset values are omitted and the transcription payload remains model-only

Risk

Low. Existing callers that do not set language or prompt produce the same payload as before. The behavior changes only for settings that callers explicitly requested but that the streamed path previously ignored.

Issue number

None. Found while auditing parity between static and streamed STT settings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5121a8b839

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Note

You have reached your Codex usage limits. You can see your limits in the Codex usage dashboard.

[chatgpt-codex-connector]

Note

You have reached your Codex usage limits. You can see your limits in the Codex usage dashboard.

[seratch]

Thanks for the contribution. Forwarding language and prompt in streamed STT is the correct direction, but the current model split is incomplete. Both gpt-transcribe and gpt-live-transcribe use the plural languages field; existing transcription models continue to use singular language.

Please apply the plural mapping to both current models and add regression coverage for gpt-transcribe, while retaining the existing singular-field and unset-value tests.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2f485f1553

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[sylvesterkaczmarek]

Addressed. Both gpt-transcribe and gpt-live-transcribe now map STTModelSettings.language to plural languages: [value], while existing transcription models retain singular language. Added focused gpt-transcribe regression coverage while retaining the existing singular-field and unset-value tests. Ready for re-review.

[seratch]

@sylvesterkaczmarek Can you complete these open PRs before sending further PRs? Thanks for your contributions!

[sylvesterkaczmarek]

@sylvesterkaczmarek Can you complete these open PRs before sending further PRs? Thanks for your contributions!

Understood. I'll focus on completing the current open PRs before sending anything new.

I also rechecked the language mapping against the current Realtime guidance and corrected gpt-transcribe back to singular language; only gpt-live-transcribe uses plural languages. The regression coverage and PR description are updated.

[seratch]

Thanks for working on this. #4574 will resolve the issue.