WhisText

To keep this service free and sustainable, we use speech recognition models with lower computational requirements. Depending on the audio quality and content, transcription results may contain errors. Please use this service with an understanding of these limitations.

WhisText — High-Accuracy Transcription & AI Summary

■Key Features ・Automatic model selection based on audio language 　・Japanese: ReazonSpeech 　・English: NVIDIA NeMo 　・Other languages or mixed-language audio: Multilingual model ・AI-powered summaries and meeting minutes (LLM) ・Server-side processing to save battery life ・Privacy-first: Audio and text data are purged immediately after processing

■Recording & Import ・One-tap recording to automatic text conversion ・Audio file import (supports Share Sheet) ・Extract and compress audio from video files

■Smart Features ・Text highlighting synced with audio playback ・Timeline markers (adjustable in 5-second increments) ・Customizable AI prompts for summaries and minutes ・Image-to-text OCR (Local processing, no additional costs)

■Export & Integration ・Export as Text, Markdown, or Subtitles (VTT/SRT) ・Automatic Slack integration ・Usage cost tracking in Settings

■Privacy & Security ・Audio files and transcriptions are permanently deleted from our servers as soon as processing is complete. We do not store any of your data.

■Supported Formats ・Audio: mp3, m4a, aac, wav, flac, ogg, opus, wma, amr, mpga, webm, aiff, caf ・Video: mp4, mov, m4v, webm, mkv, avi, mpeg, mpg