Speech Recognition whisper_1 uses OpenAI's whisper-1 model. Besides the transcript, it gives the actual start and end time of every sentence and every word; the timeline comes from the model while it recognizes the speech, which suits precise subtitles and cutting a video by its words. Just upload an audio or video file; the system converts it to mono audio and sends it to the cloud, and recordings longer than 40 minutes are split at pauses between passages and merged back on the original timeline. With SRT output, subtitles are split at sentence-ending punctuation into one sentence per entry, timed to where it was actually spoken, ready for Premiere, DaVinci Resolve, YouTube and other editing or publishing platforms. When the language is set to Chinese, results are converted to Traditional characters. Common formats (mp3, wav, m4a, ogg, mp4, mov, webm) are supported with no pre-conversion needed. If word-for-word accuracy of the transcript matters most, use "Speech Recognition gpt_transcribe"; for multiple speakers use "Multi-Speaker Speech Recognition". Maximum total upload size per request: 512MB (all files and message content combined). Output content formats: Text Pricing: Medium This module does not support continuous conversations. AI can make mistakes. Please verify important information.