Driftwatch

OpenAI / OpenAI API /

Breaking

voice no longer accepts "shitan"

What changed

A value you may have hardcoded is now rejected.

Affects LiveInitialSessionAudioOutputParam reachable from 2 operations.

no longer accepts "shitan"

Who is affected

Code that sends OpenAI's LiveInitialSessionAudioOutputParam objects, used by 2 OpenAI operations.

Before and after

WhereBeforeAfter
LiveInitialSessionAudioOutputParam.voice"alloy", "ash", "ballad", "beacon", "bossa", "brise", "cedar", "cinder", "coral", "delta", "echo", "flitz", "gleam", "harema", "juni", "marin", "meridian", "nira", "noeul", "nuri", "quartz", "ripple", "sage", "shimmer", "shitan", "sillage", "stone", "tempo", "verse", "vesper", "willow""alloy", "ash", "ballad", "beacon", "bossa", "brise", "cedar", "cinder", "coral", "delta", "echo", "flitz", "gleam", "harema", "juni", "marin", "meridian", "nira", "noeul", "nuri", "quartz", "ripple", "sage", "shida", "shimmer", "sillage", "stone", "tempo", "verse", "vesper", "willow"

Code that breaks

// sending "shitan" for voice is now rejected

Based on OpenAI’s published API specification. How severity is decided.