Toutes les versions de FastSub AI, les plus récentes en premier. Il manque quelque chose ? Envoyez une demande depuis la page d'assistance.
Version 0.10.0Dernière
AmélioréNew name and logo: FastSub AI. Same promise: 100% on your device.
NouveauVocal remover: split any song or video into vocals and a karaoke instrumental, with an AI model running in your browser.
AmélioréLong recordings use about 30% less memory: audio is decoded straight into one buffer that the AI workers share instead of copying.
NouveauTranslate many subtitle files at once and download them as one ZIP.
Version 0.9.0
AmélioréSubtitle translation works on whole sentences, then splits them back over the original lines: far fewer broken or garbled fragments.
NouveauAuto-sync can now match the spoken words to the subtitle text: accurate even with music, explosions or cut scenes (subtitles in the spoken language).
NouveauDual subtitles: combine two languages into one file, as two lines per subtitle or top and bottom (.ass).
NouveauWorks offline: pages and tools you have opened keep working without a connection.
CorrigéLink previews (Open Graph images) no longer cut off text and now show Japanese, Chinese, Korean, Thai, Hindi and Arabic correctly.
CorrigéExported videos (dubbing, noise removal) no longer have audio 44 ms behind the picture.
CorrigéCancel now really stops AI jobs, including when pressed before a step starts; auto-sync got a Cancel button.
AmélioréScreen readers announce progress on every tool, and editor line buttons have clear labels.
Version 0.8.0
AmélioréTranscription no longer invents text over music, noise or silence: a voice detector picks the parts with speech and repeated loops are removed.
NouveauNames and terms field: list people, brands or jargon and the AI spells them correctly.
CorrigéAutomatic language detection now really detects the spoken language instead of assuming English.
NouveauTikTok captions: auto-transcribe and burn animated word-by-word captions (pop and highlighted-word styles), plus karaoke .ass export.
NouveauRemove background noise from recordings and videos (fans, traffic, wind, keyboard), with before/after preview.
NouveauNew look based on Google Material Design 3 (Google blue, Roboto, standard components), a wider layout for large screens, and dark mode that follows your system setting.
AmélioréFirst visit opens in your language, guessed from your browser and time zone (nothing is sent anywhere), with a one-click switch.
NouveauSupport the project page: give by card, PayPal, Buy Me a Coffee, Ko-fi, crypto (with QR codes) or bank transfer, plus other ways to help. Help and the ticket form now live on the contact page.
NouveauSpeech to Text and MP3 to Text pages.
Version 0.7.0
NouveauInterface in 19 languages, with flags and right-to-left support for Arabic.
NouveauAI auto-sync: a neural voice detector lines subtitles up with the speech, including speed drift.
NouveauSpeaker labels for interviews and meetings, computed in the browser.
NouveauText to speech, MP3 export, LRC lyrics editor and SDH tag remover.
NouveauCapCut project import and Final Cut Pro XML export.
NouveauTranscription reuses subtitle tracks already embedded in the video.
AmélioréNew navigation, footer, tool icons, how-it-works and FAQ on every tool page.
AmélioréOpen Graph images, structured data and multilingual sitemap for better search results.
NouveauSupport tickets by email and this changelog.
Version 0.6.0
AmélioréTranslation now runs 100% in the browser: Chrome's built-in translator or Opus-MT models. No server at all.
NouveauAutomatic source language detection.
Version 0.5.0
NouveauOCR for Blu-ray (PGS/SUP) and DVD (IDX/SUB) image subtitles.
NouveauAI voiceover and video dubbing; one-click video translator.
Version 0.4.0
NouveauSubtitle translator and burn-in with classic, boxed and word-by-word styles.
Version 0.3.0
NouveauWhisper speech recognition in the browser (WebGPU or CPU).
NouveauSubtitle editor with waveform, video preview, undo and autosave.