Whisperkit vs Codictate

Whisper turns speech into accurate text — entirely on your device. Powered by Apple's Neural Engine and an open-source multilingual speech recognition model, Whisper transcribes audio recordings, voice memos, lectures, meetings, podcasts and videos directly on your iPhone. Nothing is uploaded. Nothing is tracked. No accounts. New: make music (beta). Describe a song — style, mood, instruments, even lyrics — and Whisper composes it on your device. KEY FEATURES • On-device transcription — your audio never leaves your iPhone • Record audio directly in the app, import files from Files, iCloud or Photos, share audio from other apps, or paste a link to an audio or video file • Supports many languages including English, Chinese (Simplified or Traditional), Spanish, French, German, Japanese, Korean, Russian, Portuguese, and Italian — plus auto-detection • Real-time progress with live segment-by-segment streaming • Multiple model sizes — pick speed or accuracy to match your needs • Performance monitor shows CPU, memory and Neural Engine usage in real time • Works offline once a model is downloaded • Keeps working in the background while you use other apps • Copy or share transcripts with one tap • Beautiful, modern interface designed for iOS and iPadOS MAKE MUSIC (BETA) • Describe the music and Whisper composes it on your device • Add lyrics to have them sung, in many languages • Fuller arrangement turns a short idea into a full band • Every song is kept with its prompt and lyrics, to play, share or make again WHY ON-DEVICE MATTERS Most transcription apps send your audio to a server. Whisper does not. Your meetings, voice notes, interviews and personal recordings stay private because everything runs locally on your iPhone's Neural Engine. • No accounts to create • No subscriptions, no ads • No analytics or tracking SDKs • Works in airplane mode after the first model download • Safe to use with confidential and sensitive recordings PERFECT FOR • Journalists transcribing interviews • Students capturing lectures • Professionals transcribing meetings • Podcasters generating show notes • Writers dictating drafts • Anyone who values privacy Whisper is open speech recognition done right — fast, accurate, and yours alone.

Press a button, speak, and your words are transcribed and pasted into whatever app you're using. No internet required. No account. No data leaves your phone. Codictate is AI dictation that runs entirely on your device. The AI models run locally on your iPhone's Neural Engine and CPU, so your voice is never used to train AI models. HOW IT WORKS Codictate installs a system keyboard with a Dictate button. Tap it, speak, and on-device AI transcribes your speech in real time. The transcript is then inserted into your current text field. Two AI speech engines are included: - Parakeet TDT v3: Runs on the Neural Engine. Fast, lightweight, 95.7% accuracy on clean English audio. - Whisper base (quantized): Runs on CPU as a fallback. Multilingual support with a 57 MB model. You pick which AI model fits your needs. Switch anytime in settings. PRIVACY Everything runs on-device. Audio is never uploaded. There is no server, no analytics, no telemetry. Your recordings stay on your phone and are discarded after transcription. AI dictation without the cloud. FEATURES - On-device AI transcription (no internet needed) - System keyboard with one-tap dictation - Parakeet TDT v3 (Neural Engine, fast) - Whisper base-q5_1 (CPU, multilingual) - Works in any app that accepts keyboard input - Live Activity indicator while recording - Action Button and Shortcuts support REQUIREMENTS - iPhone with iOS 17.0 or later - Apple Silicon (Neural Engine required for Parakeet) - Approximately 57 MB for Whisper model or 500 MB for Parakeet model