Convert WebM to M4A

A recording made in a browser — a voice message, a captured call, an audio note from a web app — comes out as WebM holding Opus. Safari will play it inside a web page, because the browser has the decoder. The moment the file is saved to the phone it stops being a web resource and becomes a file, and iOS has nothing that opens it: not Files, not Music, not Voice Memos, not the share sheet.

M4A is the other side of that line. Apple named the extension in 2004 for an MP4 container carrying sound and no picture, and it is what the phone itself produces when it records. Converting to M4A does not make the recording better; it makes it a first-class citizen on the device instead of an attachment nothing will open.

An M4A is an MP4. Identical container, identical structure, different extension chosen so that a double-click opens a music player rather than a video player. Inside it here is AAC — the codec MPEG standardised in 1997 as ISO/IEC 13818-7 and the one Apple built its entire audio catalogue on.

The container can hold Apple Lossless as well, which is where some confusion comes from: an M4A downloaded from one place may be lossless and one from another may not be, with nothing in the name to tell them apart. What is written here is AAC, because that is what is playable on every device that reads the container and what a browser can encode.

AAC is straightforwardly the better codec. At the same file size it holds up better than MP3, particularly on speech and on anything with cymbals or reverb, and it is the default for a reason across streaming, broadcast and Apple’s own products. If the recording is heading for a phone or a computer made this decade, M4A is the right answer.

MP3 wins in exactly one dimension and it is a big one: it plays on everything ever built. Car head units from 2006, gym treadmills, hotel alarm clocks, cheap voice recorders, the ancient laptop in the meeting room. If you do not know where the file is going, or if it is going to someone else, the worse codec in the more universal wrapper is the safer choice.

The awkward part of this problem is usually geography: the WebM is already on the phone, and the obvious fix involves moving it to a computer, converting it, and moving it back. That is three steps too many for a two-megabyte voice note.

This page runs in mobile Safari the same way it runs on a desktop. Choose the file from Files, let it convert, and save the M4A back — the decoding and encoding both happen on the phone, using the same codecs it already uses to play audio. Nothing is installed, nothing is uploaded, and the recording never leaves the device it was made on.

Opus in, AAC out, with a full decode between them. Both formats are lossy, so this is compression applied to audio that has already been compressed once, and a little more detail goes each time. There is no way around it: no browser can move audio between two different lossy codecs without decoding it first.

For speech, which is what nearly every WebM voice note contains, the result is indistinguishable at any of the three quality settings. For music the second pass is measurable and, on good headphones, occasionally audible in cymbals and room reverb. If the WebM is a rehearsal recording or a live capture, use the high band and keep the original alongside the M4A.

The M4A arrives with audio and no metadata, because there was none in the WebM worth carrying — browser recorders and meeting tools name files by timestamp and write nothing else. Dropped into Music or iTunes as-is, the recording appears under its filename with no artist and no album, which is how a library quietly becomes unusable.

Set the tags deliberately as part of filing it. Music can edit them in place, and any tagger will do it in bulk for a batch. It is thirty seconds per file and it is the difference between a recording you can find in two years and one you cannot.

Only the first audio track crosses over, and any video track is discarded entirely. A recorded meeting with separate microphone and system-audio tracks gives you whichever the container lists first, with no way to choose between them here.

Chapters and any embedded subtitle track are dropped too. For a podcast episode assembled elsewhere that is worth knowing before this becomes the delivery step: chapter marks belong to the tool that produced them, and adding them back is a job for a podcast publishing workflow rather than a converter.

Expect the M4A to be somewhat larger than the WebM for the same recording, and often noticeably so on speech. Opus is unusually good at voice at low bitrates — it was designed for real-time conversation — while AAC is tuned for music and general delivery. A five-minute voice note that was 1.5 MB as Opus can land near 5 MB as AAC at the balanced setting.

The three quality bands land at 96, 128 and 192 kilobits a second, so they move that figure down by a quarter or up by a half rather than by any factor of two. The small setting is right for anything that is purely somebody talking and is going onto a phone; the high setting is worth it only when there is music in the recording, because this is already a second generation of compression and a lean one has nothing left to give.

Up to a hundred files at once, each converting independently with its own progress, and a single ZIP when more than one has finished. Nothing queues behind anyone else because there is no shared server to queue on.

For this pair the local approach is as much about the phone as about privacy. Uploading a folder of voice notes over a mobile connection and downloading them back is slow and eats an allowance; converting them in the tab costs battery and nothing else, and the recordings — which are, after all, people talking — stay where they were.