peeklessAUD
Tools
All Peekless tools →

Audio editing help

Peekless Audio converts audio to MP3, M4A, WAV, OGG or FLAC, normalizes the volume, removes noise and silence, merges several files into one, edits tags, and in the editor it cuts audio, mixes tracks with music under speech, records from the microphone, transcribes speech to text and makes ringtones. All right in your browser – your audio is never uploaded.

Getting started

  1. Add files: drag them into the window, or click Choose files. You can add a video too – its audio is used. To make a recording, use the Record from microphone button.
  2. Pick an edit in the Convert, Clean up, Silence or Tags tab and set it up. It applies to all files in the list. To edit a single file, use the Edit button – the editor opens.
  3. Process them with the button below. Files are processed one after another; you’ll see the progress of each one and then the result.
  4. Download files one by one, or all at once in a ZIP. On a phone you can share them directly.

If you change the settings, just process the files again. Your browser remembers the settings for next time.

Bulk edits

  • Convert – the format MP3, M4A (AAC), WAV (16 or 24-bit), OGG (Opus) or FLAC, the quality (bitrate), mono or stereo and the sample rate. MP3 and M4A play on every device, WAV and FLAC are lossless, OGG is the smallest.
  • Clean up – noise reduction (wind, fans, background hum; meant for speech) in High quality (the DeepFilterNet neural network) or Fast (RNNoise) mode, with Full or Gentle strength, volume normalization to −16 LUFS like music and podcasts online, or to maximum (the loudest point at −1 dB), and fades at the start and end.
  • Silence – removes silence at the start and end, or shortens all pauses longer than the chosen length to a short breath. Raise the silence threshold if the recording contains noise.
  • Tags – removes all tags and the cover, or sets the artist, album, year and genre for all files. The audio isn’t re-encoded, so the quality stays the same.

Saved audio never goes above full volume: instead of clipping, which would sound like crackling, the loudest parts are smoothly lowered. The original’s tags are left out when converting – whatever you want to keep, enter in the Tags tab or in the editor.

Merge files

With two or more files, a Merge button appears in the list header. The files are merged in the order of the list into a new file at the end of the list – change the order with the arrows next to a file. You can mix formats; everything is converted to the sample rate of the first file. You can then process the merged file or open it in the editor, for example to add fades between the parts.

Editor: waveform and cutting

The Edit button opens the editor. At the top are playback (Space), Loop to play the selection over and over, and the time accurate to the millisecond; below are the tools and the timeline with the tracks.

  • Selection – drag across the waveform; tap to set the cursor. A selection can span several tracks.
  • Zoom – with the mouse wheel and Ctrl, a two-finger gesture, the Zoom in and Zoom out buttons, or the scrollbar below the tracks. Fit shows the whole project.
  • Cutting – Cut, Copy, Paste (at the cursor), Delete, Trim (keep only the selection), Silence (replace the selection with silence), Insert silence and Split the clip at the cursor. Snap aligns the selection and clips to the edges of other clips and the cursor.
  • Clips – drag a clip by its top bar to move it in time, even to another track. Drag the small square in the top corner of a clip to set a fade.
  • Undo and Redo go back up to 200 steps. The original file never changes; Discard changes closes the editor without saving.

Keys: Space to play and stop, Ctrl+Z to undo, Ctrl+Y to redo, Ctrl+X/C/V to cut, copy and paste, Delete to delete, S to split, Ctrl+A to select all, the arrow keys to move the cursor by a second (by 5 s with Shift), Home and End to jump to the start and end, + and − to zoom in and out, Esc to clear the selection. On a Mac, use Cmd instead of Ctrl.

Effects

Effects in the Effects panel apply to the selection, or to the whole track when nothing is selected:

  • Volume – make it louder or quieter by decibels, to maximum (−1 dB), normalize to −16 LUFS, fade in and out.
  • Clean up – reduce noise (High quality: the DeepFilterNet neural network, its 9 MB model is downloaded only the first time; Fast: RNNoise; strength Full or Gentle), remove silence with an adjustable threshold and pause length.
  • Vocals and music – Remove vocals (karaoke), Vocals only, or Split into 2 tracks (vocals and music separately; mix them to taste with the track volumes). They’re separated by the Spleeter model (40 MB, downloaded only the first time). The quality is good, not studio-grade; use it only on your own recordings or where the rights holder allows it.
  • Speed and pitch – speed from 0.5× to 2× with the voice keeping its pitch, pitch ±12 semitones without changing the tempo.
  • Equalizer – the presets Clearer voice, More bass, More treble, Phone and Radio, plus a Compressor that evens out quiet and loud passages (gentle or strong).
  • More – reverse the audio, convert to mono, swap the left and right channels.

Multiple tracks and music under speech

In the Tracks panel you can add a file to a new track, an empty track or a recording. Each track has a name, M (mute), S (solo – play only this track), volume and pan in its header. The speaker icon turns on ducking: the music track gets quieter on its own while the other tracks (speech) play, and comes back up in the pauses. Set the ducking depth in the Tracks panel. What you hear during playback is exactly what gets saved.

Recording from the microphone

On the home page, record a new file with the Record from microphone button; in the editor, record to a new track or at the cursor. Choose the microphone, mono or stereo, and a 3 s countdown before starting; while recording, you see the signal level. The first time, your browser asks for permission to use the microphone. The recording is saved continuously to your browser’s storage on your device, so it can be long – and it’s never sent anywhere.

Speech transcription and text-based editing

The Transcript panel transcribes speech with the Whisper model right on your device. Choose the spoken language (or Automatic) and the quality: Fast (about 77 MB) or Accurate (about 249 MB, better for languages other than English). The model is downloaded from our server only the first time.

For an interview or a meeting, check Identify speakers (choose the number of speakers, or leave it on automatic). The transcript is split into “Speaker 1”, “Speaker 2”… paragraphs; tap a speaker’s name to rename them. The names also appear in the TXT and the subtitles. Speakers are identified by the pyannote and CAM++ models (about 35 MB, downloaded only the first time) – they can mix up similar voices.

Tap a word to jump to its place in the audio. Drag to select words, and the Delete selected words from the audio button cuts them out of the audio too. A single button removes all filler words (“um”, “uh”). You can search the transcript and download it as TXT, or as SRT or VTT subtitles. The transcript may contain mistakes – check it before you use it.

Chapters

The Chapter button on the toolbar (or the M key) adds a chapter at the cursor. Drag it elsewhere in the ruler, or double-click it to rename it; tap it to jump to its start. When you cut, the chapters move with the audio. The Save panel has a list of chapters where you can rename or delete them, copy them as text (“00:00 Intro” – for an episode description or a YouTube video, for example), paste them from text, or download them as JSON for podcasts (Podcasting 2.0).

Chapters are written to MP3 (ID3 – Apple Podcasts, Overcast, Pocket Casts and others), M4A (VLC, foobar2000 and other players), OGG and FLAC files. WAV doesn’t support chapters.

Saving, ringtones and tags

In the Save panel you choose the format, quality, channels, sample rate and range (the whole project or just the selection). Save tags writes the title, artist, album, year, genre and album cover (JPG or PNG); without it, the file is saved without tags. Done takes your edits back to the main screen – the file is then processed with them.

iPhone ringtone saves the selection (40 s at most) as M4R with fades at the start and end. Transfer it to your iPhone with Finder on a Mac, or with the Apple Devices app or iTunes on Windows; without a computer, add it to a GarageBand project on your iPhone and choose Share → Ringtone. For Android, save the selection as MP3.

Formats, browsers and speed

We can open MP3, M4A and AAC, WAV, OGG and Opus, FLAC, WebM and the audio from MP4, MOV, MKV and 3GP videos. Your browser decodes and encodes the audio (the WebCodecs API; the MP3, AAC and FLAC encoders run in WebAssembly). An hour-long recording usually converts in anywhere from tens of seconds to a few minutes.

It works in current Chrome, Edge, Safari and Firefox. If your browser can’t handle a format, we tell you up front – files are never sent to a server “just in case”.

The editor keeps audio in your browser’s storage on your device, not in memory, so it can handle long recordings too. That’s why it needs free space of roughly 10 MB per minute of stereo audio; the space is freed when you close the page.

Work in progress

Your files, recordings and edits in the editor are continuously saved, encrypted, in your browser. When you reload the page, or when your phone closes the tab in the background, everything comes back on its own – including the open editor and an unfinished recording. After you close the tab, you’ll find your work on the home page in the In progress in this browser list (restore it with its button, delete it with the cross). Files up to 100 MB are saved; add larger ones again after restoring. Everything is deleted 7 days after the last edit. See the privacy policy for details.

App and offline use

You can install Peekless Audio as an app – with the Install button in the header (Chrome, Edge, Android) or in Safari with Add to Home Screen. After the first use it works without an internet connection too (except for the first download of the models for transcription, high-quality noise reduction and vocal separation; without them, noise is reduced with the fast method). On Android you can then share audio files to the app from other apps.

Privacy

Neither your audio nor your recordings are ever uploaded – your browser processes them and the results stay with you. We use no cookies and no tracking. See the privacy policy for details.

Libraries and licenses

Peekless Audio uses these free libraries (they run in your browser and are downloaded from our server only when needed):

  • Mediabunny (MPL-2.0) – reading and writing MP3, MP4/M4A, WAV, OGG, FLAC and WebM.
  • The AAC audio encoder from FFmpeg (LGPL-2.1, the @mediabunny/aac-encoder package under MPL-2.0; only in browsers that can’t encode AAC themselves), the LAME MP3 encoder (LGPL-2.0, the @mediabunny/mp3-encoder package under MPL-2.0) and the FLAC encoder from libFLAC (BSD-3-Clause, the @mediabunny/flac-encoder package under MPL-2.0); source and license.
  • RNNoise noise reduction (BSD-3-Clause, the @shiguredo/rnnoise-wasm build under Apache-2.0).
  • Speech transcription: transformers.js (Apache-2.0), ONNX Runtime Web (MIT) and the Whisper model by OpenAI (MIT) – downloaded only when you create a transcript.
  • Models run by ONNX Runtime Web (MIT), downloaded only when you use the feature: DeepFilterNet3 noise reduction (MIT / Apache-2.0), Spleeter vocal separation by Deezer (MIT), and for speakers pyannote segmentation 3.0 (MIT) and 3D-Speaker CAM++ (Apache-2.0); ONNX conversions from the sherpa-onnx project (Apache-2.0).
  • Lucide icons (ISC), Inter and Outfit fonts (OFL-1.1).