Audio

The audio editor is a multitrack editor in the page. Open WAV, MP3, AAC, Ogg or FLAC - the 8 kHz recordings telephone systems make included - and arrange them as clips on tracks: cut, copy, paste, split, trim, move, fade in and out with four curve shapes, change the volume of a stretch, play backwards. Nothing touches the original recordings, and every step undoes. Mute, solo, volume and pan per track; zoom from the whole recording to single samples; a frequency view for any track; markers and regions.

Loudness is measured to the broadcast standards and set to the target for streaming, podcasts, speech or broadcast; silence is found and removed. And it is built for call recordings: give it the transcript a speech-to-text service returns and it finds card numbers, security codes, bank details, phone numbers, email addresses, postcodes and dates of birth however they were said - and card numbers keyed in on the phone's keypad - and bleeps them on every track, keeping a log of what was removed. Audio never leaves the page.

Redacting a card payment call

A made-up customer call, generated in the page: the agent on one track, the caller on the other, with a transcript. Find the personal data, bleep it, play it back, export the redacted call and the log. The voices are synthetic babble, so listen for the bleeps rather than the words.

Find personal data

Bleep everything found

Measure loudness

Export the redacted call

Keyboard: with the timeline focused, Space plays and pauses, the arrow keys move the playhead (Shift selects), Up and Down change track, S splits, Delete deletes, B bleeps, M adds a marker, I and O set the selection, Ctrl+Z undoes.

// Node 18 or later, from the Gold package: bleep every call recorded today before it is archived.
import { decodeWav, encodeWav, readWords, findRedactions, singleClipProject, addRedactions, renderProject } from './dist/node/bmx-audio-editor.mjs';

for (const call of await todaysCalls()) {
  const audio = decodeWav(await readFile(call.wav));
  const words = readWords(await readFile(call.transcript, 'utf8'));
  const found = findRedactions({ words, samples: audio.channels[0], sampleRate: audio.sampleRate });
  const project = addRedactions(singleClipProject(audio, 'call'), found).project;
  await writeFile(call.archivePath, encodeWav(renderProject(project, new Map([['call', audio]]))));
  await log(call.id, found.map(f => ({ start: f.start, end: f.end, what: f.label })));
}

Tidying a recorded talk

A talk recorded from the back of a room: long pauses, and far too quiet. Find the pauses, take them out, and bring the level up to the podcast target - each step undoes.

Find the long pauses

Take them out

Bring it to -16 LUFS

Show frequencies

Get the code

<script src="/assets/bmx-components.min.js"></script>

<bmx-audio-editor id="au-talk" label="Recorded talk" style="inline-size: 100%; --bmx-audio-editor-height: 18rem"></bmx-audio-editor>