Multi-Speaker Speech Recognition is powered by OpenAI's cloud gpt-4o-transcribe-diarize model, the diarization-enabled variant of the gpt-4o-transcribe family. The transcript automatically labels each speaker (SPEAKER_01, SPEAKER_02, ...), so meeting minutes, interviews, panels, and customer-support calls come back ready to read. Just upload an audio or video file (mp3, wav, m4a, ogg, mp4, mov, webm - no pre-conversion needed); the audio is split into 30-minute segments and speaker labels stay consistent within each segment; the transcript streams in progressively while recognition runs, then the results are merged into a single plain-text transcript. Recordings longer than 30 minutes span multiple segments, and the output marks each split point with a note that speakers before and after it may not match; for important recordings, uploading in 30-minute parts is recommended. It keeps the strong noise robustness, mixed-language handling, proper-noun accuracy, and well-preserved punctuation the gpt-4o-transcribe family is known for. SRT subtitle output is not supported by this module; for single-speaker recordings or SRT subtitles, use "Speech Recognition" instead. Maximum total upload size per request: 256MB (all files and message content combined). Output content formats: Text This module does not support continuous conversations.