Doesn’t work
Response from developer
Try the new version. If it still doesn't work, please send me an email at liutao124@gmail.com

This app converts voice recordings into text and extracts key information. It offers offline transcription, speaker separation, and AI-powered summaries, with all processing done on your device for enhanced privacy. Export your transcribed notes in various formats.
1.13.0
November 15, 2026
What's New 1. Export Reports to Word (headline feature) - Export any generated report as a .docx file. The button appears only when the currently selected template has a finished report - no dead taps on templates that are empty or regenerating. - Proper heading levels (so a table of contents works), well-proportioned tables and page margins: Chinese text reads cleanly in Word and WPS. - Decisions, action items, interview quotes, study points and timelines are laid out by template, with the matching recording timestamps next to key findings for easy verification. 2. Reports and recordings survive reinstalls and interruptions - Reports you already generated no longer disappear after reinstalling the app, including reports from your own custom templates. - If a recording is interrupted or the app quits mid-session, the audio stays playable; the app repairs it automatically on next launch instead of showing an unsupported-format error. - Recovery no longer creates duplicate broken entries, so your list stays clean and no needless re-transcription is triggered. 3. A smoother generation flow - You're told right away that the task is processing in the background - keep recording and just refresh later. - Removed the spinning overlay over the preview; the button state resets correctly once generation finishes. - Queued tasks now show the correct names and statuses. - A failed generation shows one actionable message instead of two contradictory pop-ups. Fixes & Improvements 1. Fixed reports failing to parse when the model produced long output (meeting / interview / class templates). 2. Fixed the "Generating..." button state sticking around after a report had already finished. 3. Fixed reports or transcripts occasionally not loading after re-transcribing or moving to a new install. 4. Fixed a queued task showing a raw untranslated internal label. 5. Updated the diagnostics entry path in both English and Chinese Privacy Policies. 6. Synced UI copy in all five languages (Simplified Chinese / English / Japanese / Korean / Cantonese). Notes - After your trial ends, you can still record 1 item free every day; the allowance resets automatically the next day. You can also extend the trial once by 7 days. - Pay once, use forever: no subscription, no ads - one purchase unlocks everything. - Live captions and same-engine transcription remain exclusive to iOS 26; earlier versions can download the high-quality recognition models (~200MB) with all other features unaffected. - The on-device AI model (~2.55GB) is downloaded once in Settings for offline report generation. All local recognition, transcription and AI processing run entirely on your device; audio never leaves your phone. Only when you actively use "AI Assistant" will the transcript be sent to the app you choose, and exported reports leave the device only when you choose to share them.
More$1.99
VoxSafe Pro lifetime unlock
Lifetime VIP, all features, one-time purchase.
Try the new version. If it still doesn't work, please send me an email at liutao124@gmail.com
No, VoxSafe is ad-free. The app is free to download and offers a 7-day full-featured trial. After the trial, you can still record once per day for free.
VoxSafe supports offline transcription for Chinese, Cantonese, English, Japanese, Korean, and more. This functionality does not require an internet connection.
VoxSafe is free to download and includes a 7-day trial of all features. A one-time purchase of $1.99 unlocks all features permanently, with no subscriptions or hidden charges.
VoxSafe is available on iPhone and iPad devices. It supports background recording and has Home Screen & Lock Screen widgets for instant recording.
The latest version of VoxSafe is 1.13.0, last updated on November 15, 2026. Updates typically include new features, bug fixes, and performance improvements.
Yes, VoxSafe automatically separates voices and can recognize named speakers in future recordings. This feature helps in organizing and understanding multi-speaker audio.
Transcripts can be exported in various formats including PDF, Word, Markdown, TXT, SRT, and CSV. Audio can also be exported in WAV or M4A formats.
Yes, VoxSafe prioritizes privacy by performing all transcription, recognition, and AI minute generation on your device. Your audio never leaves your phone, and the app includes Face ID/passcode lock for sensitive content.