What's new at Formavide

Everything user-facing we ship, month by month — new tools, engine upgrades, fixes, and the occasional removal. Newest first. We only write entries when there's something you'd actually notice; quiet months stay quiet.

  • Fixed

    A VOCALOID5 project file (.vpr) dropped into the music converter now gets told what it is — a project file, not an encrypted track — instead of a dead-end decryption error.

  • Improved

    AI tools that download a model on first use — subtitles, vocal separation, noise removal — now say so before starting, download size included. Once the model is cached in your browser, the note disappears.

  • Improved

    Homepage rebuilt around how people actually arrive: the AI tools — noise removal, vocal splitting, subtitles — now lead the tools grid, the full overview sits right below it, and the most-used guides are linked at the bottom of the tools section. All tools →

  • New

    Guides now ship weekly. Recent reads: why DIY acapellas sound shaky, why free online cutters stamp watermarks, TS vs MP4, and how to trim a large video without uploading it. Read the blog →

  • Fixed

    Files from newer KuGou versions — often disguised as .flac but actually KuGou's new per-file encryption — now get a clear explanation of why they can't be decrypted offline, plus the way out (re-download in another quality or convert inside the KuGou app), instead of a generic parse error.

  • Fixed

    The 2-stem vocal separation engine crashed on start for GPU-accelerated browsers (Chrome, Edge, Firefox on Windows) — it had been unintentionally running on a broken WebGPU path. It's back on the universally compatible one, and both engines now report which one they used.

  • New

    Trim went live — frame-accurate video cutting with input-side seeking, re-encoded from your exact cut points. Files up to 1.5 GB on desktop. Cut a video →

  • Improved

    Subtitles: soft-embed arrived — SRT goes into an MP4 without re-encoding, done in seconds — and hard-subtitle burning moved to a streaming pipeline, so long videos export in a fixed memory budget. Try subtitles →

  • New

    The blog launched with six guides — containers, podcast noise, large-file trimming, audio extraction, karaoke tracks, subtitle types — and hasn't missed a weekly cadence since. Read the blog →

  • New

    Sign in with Google. One tap covers login and sign-up; email with 6-digit codes still works the same as before. Sign in →

  • Improved

    The free daily quota now shows on every tool page before you load a file — no surprises after you've committed.

  • Improved

    Every tool moved onto one shared layout: the work area on the left, tips and FAQ on the right, the same shape across all of them.

  • New

    License plans went live — monthly, annual, and a one-time lifetime option. Each tier's limits are spelled out on the pricing page before you buy. See pricing →

  • New

    Nine dedicated converter pages for single-step jobs — WAV to OGG, M4A to MP3 and their neighbors — each with its own how-to and FAQ. Browse tools →

  • New

    New brand mark: an FV interlock drawn from Format + Video, replacing the old letter-and-triangle.

  • New

    Noise removal went fully local: DeepFilterNet3 on ONNX Runtime Web, running in your browser with adjustable strength — nothing uploaded. Try the denoiser →

  • Fixed

    Encrypted-music conversion hardened: decrypted QMC output now gets magic-number validation before encoding, so broken mflac conversions no longer slip through as garbage files.

  • New

    AI subtitles went fully local — Whisper transcription in your browser (WebGPU when available, WASM fallback) plus hard-subtitle burn-in for MP4 export. Try subtitles →

  • Fixed

    Decryption for the QMC family (mflac, mgg and friends) got the complete algorithm, replacing a simplified version that failed on some files — encrypted-music conversion stopped being a coin flip.

  • New

    OPUS and AIFF joined the audio output formats, completing the converter's seven-format audio set.

  • New

    Vocal separation went live: Spleeter running locally in your browser — vocal and instrumental in one pass, MP3/WAV/FLAC/OGG output. Split a song →

  • Improved

    Watermark rendering moved to a segment-streaming pipeline — long videos process in a fixed memory budget instead of loading the whole track into buffers. Add a watermark →

  • New

    Formavide went live: a video and audio converter running FFmpeg compiled to WebAssembly — MP4, MKV, MOV, AVI in, H.264 and H.265 encoding, everything processed in the browser tab. Convert a file →

  • New

    Watermark tool arrived — image and text watermarks burned in with a WebCodecs-accelerated pipeline. Add a watermark →

  • New

    Container-only conversions (MOV/AVI/MKV to MP4) learned stream copy: when no re-encode is needed, the streams are copied untouched — bit-identical output, near-instant. Convert a file →

Formavide processes every file on your device — nothing is ever uploaded. See the privacy policy for what that means in practice, or browse all tools.