Every change to ChunkAudio that you can see or use, newest first. Files still never leave your browser.
Loudness per part, and all parts at one level
Measure loudness, free: every part gets its integrated loudness in LUFS and its true peak in dBTP, measured on your device the way podcast apps and streaming services measure it (ITU-R BS.1770). The results list shows how far apart the parts sit.
Match loudness, Pro: bring every part to −16 LUFS for podcasts, −14 for streaming and YouTube or −23 for broadcast. Each part is raised or lowered as a whole, nothing is compressed, and no part is pushed past −1 dBTP; one that would clip stops short and the list says so. WAV and MP3 downloads carry the matched audio; an undo brings the originals back.
A page of its own:the loudness normalizer, with a link that takes one file as a whole and measures it at once, no split needed.
Huge parts say what they weigh. A part over 512 MB as WAV now shows its MP3 size next to it, and a WAV that cannot be written to a synced folder tells you so, with a way to try again in a local folder.
A yearly plan and a team plan, and the lifetime price changes on 1 November
Pro by the year: €29. For people who split every week: one payment, valid for a year, no automatic renewal; the app reminds you two weeks before the end. The 7-day pass (€4.99) stays for the single job.
Teams: five seats for €79 a year. One key for an agency, a transcription office or a newsroom, with room for the devices five people use. One payment, no automatic renewal.
The lifetime licence moves from €24 to €49 on 1 November. Until 31 October it stays at the founder price, and every licence bought by then keeps working as it is, for good. Every plan is one payment; nothing renews by itself.
Know when a job is done, even from another tab
The tab tells you. While a split, a cut, a pause removal or a download runs in a background tab, the tab's title shows the percent. When the job is done, the title and the tab icon say so, and the installed app shows a badge, until you look at the tab again. Nothing to allow, nothing to set up.
The progress bar keeps pace with the work. Each step of a job now owns the share of the bar it takes in practice (on a decoded file about half for the decode and most of the rest for writing the parts), the bar never runs backwards, and the decode estimate follows the length of the audio instead of the size of the file. The last few percent no longer take half the time.
A notification, if you want one. When a job takes more than a few seconds, the progress card offers "Tell me when it's done". Say yes once, and from then on a job that finishes while you are in another tab or window posts a system notification; a click on it brings you back. One click turns it off again. ChunkAudio never asks for that permission on its own, and the choice stays on your device.
Apple Lossless in every browser, short smart-split parts on FLAC and OGG, and plainer words for three kinds of file
Ogg voice notes and recordings (Opus, Vorbis) are split by copying. Exports from WhatsApp, Telegram and Signal and .opus recordings of one or two hours used to be decoded as a whole, which ran phones out of memory. The split now copies whole Ogg pages, so there is no length limit, the free WAV is decoded part by part, and with Pro every part stays OGG, untouched. Opus parts begin one page early with the pre-skip raised by that page, so the cut is seamless.
Apple Lossless (ALAC) now works in Chrome, Edge and Firefox too. Those browsers have no ALAC decoder, so the WAV download, the preview and the pause search used to need Safari. ChunkAudio now brings its own decoder, fetched the first time an ALAC file needs it (about 20 KB) and run on your device like everything else. Splitting ALAC was always a copy and stays one.
Smart split with short parts on FLAC, OGG and the other formats that are decoded could pick the same pause twice and stop with an error. The search now always moves forward from the previous cut, as it already did for MP3, M4A and WAV.
Three kinds of file get a plain explanation instead of "could not load": a WebM or Matroska file (usually a video; extract the sound first), a Dolby Digital (AC-3) or DTS stream (browsers do not decode them), and a file that begins with empty data, which is a download or export that was not finished.
The "what's next" cards under a result follow the job. Parts sized for a transcription service (15 minutes or 25 MB and under, or split by file size) get two transcription services, Happy Scribe and Rev, and the chunk-size guide; longer parts the podcast tools; a cut section the ringtone guide when it is short and the clips guide when it is not; a pause removal the clean-up tools. Happy Scribe and Rev are plain links, not partner links.
Chapter marks, presets, remembered settings, and a sample when a file fails
Chapter marks after a split
The cut points as a chapter list under the parts: "00:00 Part 1, 15:00 Part 2, ..." in the form YouTube descriptions and podcast show notes take. Change the names, then copy or download the list. Free.
With Pro, the same marks as a CUE sheet (foobar2000, VLC, CD burning and tagging tools) or a Podcasting 2.0 chapters JSON for your podcast host.
Presets, and settings that stay
Your last settings come back. Part length, unit and the tab you used are remembered on this device, so a weekly job needs no typing. The MP3 quality was remembered already.
Presets above the length field. Two built-in ones for everyone, "Transcription · 25 MB" and "Chapters · 60 min". With Pro, save your own: a name for the length, unit and MP3 quality you use, applied with one click, kept on this device.
Help fix an error with a sample
When a file fails to load or convert, the app now offers to send a sample: the first 256 KB and the last 64 KB of the file, the piece that failed if there is one, and the error details. No file name, and nothing is sent unless you press the button.
This is the one exception to "your files stay on your device", so it is opt-in every time, says exactly what it sends, and the sample is deleted after 30 days. The privacy policy has the details.
Several files in one go, and fewer files that fail
Several files in one go
Drop or choose several files at once. They line up under the file card, sorted by name, and Quick Split or Smart Split runs through all of them with the same length. The parts come back grouped by file, each still playable and downloadable on its own, plus one ZIP for everything with a folder per file.
The first file of a batch is free, as splitting always was. The rest of the list is a Pro feature; without Pro, "continue with the next file" takes them one at a time.
Cutting a section and removing pauses stay one file at a time.
Fewer files that fail
"Lost access to the file" on computers. Files up to a few hundred megabytes are now read into memory when chosen, on computers as on phones, so a download that is still finishing, a cloud folder that fetches on demand or a sync client touching the file no longer breaks the job halfway. This was the most common error of the past three weeks.
MP3s joined from several files (audiobooks, chapters pasted together) are measured by counting their frames, so the parts cover the whole file and not only the first chapter; the tags and cover art between the chapters are left out of the parts, which stopped the WAV conversion in Firefox.
An MP3 whose tag declares the wrong size, recordings from OPPO phones (an M4A behind a marker of the recorder app) and downloads that were cut short now load and split without decoding.
When a part cannot be decoded in this browser, the message now says so once, with the Pro same-format download as the way out, instead of decoding the whole file again for every part and ending in a memory warning.
Split by file size, and a 7-day Pro pass
Split by file size
"MB per part" next to minutes and seconds: give a size, for example 25 MB for the upload limit of a transcription service, and every part stays under it. The line below the field shows how much audio fits into one part.
The size counts for the file you download: WAV for free downloads, the original MP3 or M4A with Pro, which holds about ten times as much audio in the same size.
The feature survey under the results is gone. Thank you to everyone who answered; this feature came from one of the replies.
Pro for one job: the 7-day pass
Everything in Pro for seven days, €4.99. One payment, no subscription, and it ends by itself. The lifetime licence stays at €24 once.
The pass key works like the lifetime key on every device you use, shows its end date, and the app says when it has run out.
More files accepted, clearer messages when a file cannot work
Files with an unusual name are checked by their content. An MP3 saved as ".mpeg", a file called "talk.m4a (2)" or a file without an extension is now opened instead of turned away.
Long WAV parts download as before. Only parts above 1 GB go through the save dialog now, phones never do, and if writing to the chosen folder fails, the part is built in memory and downloaded the usual way.
Pro MP3 downloads of an MP3 source stay the original frames, also after changing the MP3 quality setting. Before, the change made long parts fail with "Unable to decode audio data".
Apple Lossless outside Safari, MP2 audio in Chrome and Edge, AMR, copy-protected tracks, videos without sound and web pages saved as audio now get one plain explanation straight away instead of several failed attempts.
Long parts as WAV, written straight to disk
No more size ceiling on the desktop. In Chrome and Edge, a part whose WAV would be larger than 256 MB is saved through the browser's save dialog and written piece by piece, so a three-hour WAV never has to fit in memory. Up to 4 GB, the WAV format's own limit.
Phones keep a copy. Files up to 150 MB are read into memory when chosen, so a tab in the background or a file from a messenger no longer ends the job with "lost access".
Files that only exist in the cloud (OneDrive, Google Drive, iCloud on demand) or are still being recorded get a message that says so.
An M4A with a stray MP3 sync word inside is no longer mistaken for an MP3.
Files from YouTube downloaders, and steadier conversions
Fragmented MP4 is split by copying now. Many "MP3" and "M4A" files from YouTube downloaders are fragmented MP4 inside. They used to need a full decode, which failed on long files; now the parts are copied frame-exact like any M4A, at any length.
One conversion at a time. Clicking several WAV downloads in a row queues them instead of decoding all at once, which ran computers with 8 GB out of memory.
A part that would become a WAV larger than a browser can build (4 GB, 1 GB on phones) is refused with a clear message before minutes of work are lost.
Raw AAC recordings that start with an encoder configuration frame no longer fail in Chrome.
MPEG Layer II files, WMA, AMR and AIFF are named as such instead of a generic "could not read" message.
Install ChunkAudio as an app
Add it to your home screen. On Android and in Chrome or Edge, an "Install app" card appears after your second job; on iPhone and iPad use Share → Add to Home Screen. It opens full screen in its own window.
Works offline. Once opened, the splitter runs without a connection. Your files never left the device anyway; now the app does not need the network either.
Share to ChunkAudio on Android: send a recording straight from your voice recorder or file manager to the installed app.
Shortcuts on the app icon: "Split into parts" and "Remove pauses".
Copies instead of decoding: cuts, pause removal, smart split, and the same format out with Pro
Cuts and pause removal keep the original format, at any length
Cut a section and Remove pauses now copy frames for MP3, M4A, MP4, MOV and AAC instead of decoding the whole file. The waveform and the pause analysis read the file a couple of minutes at a time, so a three-hour recording works, on a phone too.
Pro gets the result in the original format: a cut M4A stays M4A, an MP3 without its pauses stays MP3, untouched. Cuts land on the nearest frame, about 23 milliseconds.
Where a copy is not possible, the browser's own AAC encoder writes the M4A (Chrome, Edge and Safari on most devices); otherwise WAV and MP3 remain.
Previews of cuts and pause removal play as files, so they work with the iPhone ringer switched off.
Smart split for long files, without decoding them
Smart split now works at any length for WAV, MP3, M4A, MP4, MOV and AAC. It reads and decodes only the seconds around each planned cut to find a pause, instead of decoding the whole file. Two-hour lecture recordings that failed with "could not decode" now split in seconds, on phones too.
Long parts convert to WAV in segments, a few minutes at a time, so a 60-minute part no longer needs a gigabyte of memory.
WAV files with large metadata in front of the audio (field recorders, DAW exports) are recognised again.
Raw AAC parts preview and convert in every browser.
Clearer messages when the browser loses access to the file or runs out of memory.
Previews play on iPhones with the ringer off, the Play button shows what is playing, and the free WAV download is labelled as such.
Same format out as in, with Pro
M4A, M4B, MP4 and MOV are now split by copying the audio samples into new M4A files: no decoding, no length limit, Voice Memos and Zoom recordings of any length. Free downloads stay WAV; Pro gets the M4A parts untouched.
Raw AAC files are cut on frame boundaries like MP3.
FLAC in, FLAC out: Pro saves every part, cut or cleaned recording as FLAC, bit-exact.
Cuts and pause removal from M4A or AAC were saved as WAV or MP3 at first; since later the same day they keep the format too (see above).
MP3 output moves to Pro; WAV stays free for everything
What changed. Since 8 September, splitting an MP3 gave you MP3 parts for free. From today the free download is WAV for every source, and MP3 output is part of Pro: for MP3 sources a frame-exact copy with the original tags, for everything else a 192 or 320 kbps encode.
What did not change. MP3 files of any length still split in seconds, because the frame split stays; the WAV for each part is made from that part alone, so a seven-hour file still works. Splitting, cutting and pause removal on short files stay free.
Why. Pro is what funds the site, and the two-week experiment with free MP3 output left it with nothing to offer.
MP3 files of any length, split without re-encoding
Frame-exact MP3 splitting. MP3 files are now cut on frame boundaries by copying, the same way WAV is cut by bytes. A 7-hour recording splits in seconds and the old 2 to 3 hour memory limit is gone. (Since 20 September the MP3 download of the parts is a Pro feature; see above.) Smart split still finds silences first, then cuts the original.
Each part carries the original tags and a proper length header, so players show the right duration and can seek.
Links in three guides pointed at addresses without the .html ending and returned 404; fixed, and such addresses now redirect.
Fixed: preview after a fast split. The play button on the parts did nothing after a WAV or MP3 split that used the copy path. It works again.
Guides tidied. The lossless-splitting explainer merged into Split Audio Without Losing Quality, which now covers the frame-copy MP3 path; the voice-cloning guide rewritten around cut mode and pause removal; one article without readers removed.
Cut, remove pauses, go Pro
Cut a section. A second mode in the app: drag two handles on the waveform or type start and end times, preview the selection, then keep it or remove it. Sample-exact on WAV. Try it or read the MP3 cutter and audio cutter guides.
Trim silence. One button in cut mode places the handles at the first and last sound.
Remove pauses. A third mode finds long pauses inside a recording, shows them on the waveform, and shortens them to a length you choose. Free for recordings up to 10 minutes. Guide: silence remover.
ChunkAudio Pro. A one-time €24 licence: no ads, MP3 export at 192 or 320 kbps, pause removal on recordings of any length. Paid through Stripe, activated with a key, no account. Details.
M4A support explained. Voice Memos and Zoom recordings, what your browser can decode, and why the output is WAV: M4A splitter.
Long WAV files. WAV files of any length are split by copying bytes instead of decoding the whole file, which fixes multi-hour recordings that used to fail.
New look. Fresh design across the site, a shared header and footer, and an About page that says who runs ChunkAudio.
Small fixes. Dropped files are accepted by extension as well as type, and a rejected drop now says why.