Private browser-based audio compression

Audio Bitrate Comparison

Create two to five controlled bitrate versions from one measured source, one at a time, entirely in your browser. Switch versions near the same playback position, compare actual encoded size and average bitrate, or hide the settings in blind mode until you reveal them. No listening choice or file data is uploaded.

Choose audio file Click, press Enter, or drop one supported audio file here MP3, WAV, M4A, AAC, OGG or FLAC · maximum 500 MiB

Local processing: audio, filenames, tags and waveform data are not uploaded or added to analytics.

StatusChoose audio file

Master comparison player

Switching keeps the nearest shared playback position and clamps to the shortest measured duration; it is not sample-accurate.

Version	Actual bitrate	Actual size	Sample rate	Channels	Lowest acceptable version	Download

Why bitrate comparisons should use the same source

A useful bitrate comparison changes one encoding budget while keeping the source, output codec, sample rate, and channel policy common. Comparing unrelated files mixes recording, mastering, performance, and codec differences with the bitrate question. This tool probes one source, then serially generates every selected version using the same shared output settings.

Each output is probed independently. The table shows real bytes, measured average bitrate, sample rate, channel count, and duration rather than repeating form values. Container overhead and encoder behavior can make the measured average differ from the nominal target. Those differences are part of the comparison, not an error to hide.

Speech and music react differently

Speech, podcasts, and music expose different artifacts. Speech listeners may notice consonants, room noise, or warbling at low rates. Podcasts can include music beds and multiple speakers. Music can stress stereo image, transients, dense harmonics, and high frequencies. The page asks you to label the content type explicitly and changes suggested starting rates; it does not guess from the waveform.

Suggestions are starting points rather than quality grades. A clean mono voice may remain useful at a lower rate than a dense stereo mix, while noisy speech can be unexpectedly difficult. Use the destination device and realistic headphones or speakers. Keep the source so a low-rate delivery choice never becomes the only master.

File size vs perceived quality

For a fixed duration, a lower encoded bitrate generally means fewer output bytes. Perceived differences do not follow a universal linear score. The codec, source history, listener, playback level, environment, and content determine what is audible. Accordingly, the chart and data table use actual size and actual bitrate only; they never manufacture an audio-quality percentage.

You may mark the lowest acceptable version for your present listening setup. That note stays in the page and is not sent to analytics or a server. It is a personal planning aid, not a controlled experiment, hearing assessment, codec ranking, or claim that other listeners will choose the same version.

Listening setup and limitations

Use a quiet environment, match playback volume, and listen at several representative positions. Switching reuses a master playback time and clamps it to the shortest measured duration. Encoder delay and small duration differences mean the change is only near the same position. This implementation does not claim sample-accurate, gapless, or synchronized multi-player playback.

Browser media support can differ from the local encoder. If a valid AAC, OGG, or other result does not preview, download it or use MP3 for the listening pass. Do not increase volume only for one version. A short obvious passage can be useful, but checking the beginning, a dense section, and a quiet tail reduces accidental conclusions.

Why a higher bitrate is not always audibly better

A higher target gives the encoder more data, but the source may already be limited, previously compressed, noisy, or simple enough that the difference is not audible under current conditions. A larger number cannot recover discarded information. Re-encoding a low-rate source at 320 kbps usually makes a larger representation of the decoded source, not a restored original.

When two versions sound the same to you, choose from compatibility, size, and future use rather than assuming the largest must be best. For an archive, keep a lossless or original master. For delivery, select a tested copy. The measured result table helps separate those roles.

Re-encoding caveats

Every comparison output is newly encoded. If the source is MP3, AAC, or Vorbis, this adds another lossy generation. Comparisons can still answer a delivery-size question, but they do not represent encoding from an untouched master. For codec evaluation, begin with a WAV or FLAC source whose provenance you understand.

The engine processes versions serially and releases each virtual input/output after the result is copied. Output object URLs remain only for the active comparison and are revoked on cancellation, restart, or page exit. Estimated total output warns about workload, while actual browser memory use also includes the engine, input copy, and transient buffers.

How blind comparison works

Blind mode labels versions A, B, C, and so on. Before reveal, the chosen target rates are removed from the settings DOM and the result table, button names, and download names do not disclose them. After reveal, measured settings become visible. This reduces one obvious expectation cue but does not randomize trials, level-match codecs, or collect statistical evidence.

Normal mode displays measurements immediately. In either mode, switch at a stable passage and use the same playback chain. A blind preference can be meaningful for your task without becoming a universal finding. The tool never uploads your selected version, acceptable marker, source name, tags, or audio.

FAQ

How many bitrate versions can I create?

Choose two to five. They are encoded serially to avoid a large parallel memory peak.

Does blind mode hide answers from accessible names?

Yes. Before reveal, controls and table cells use only letter labels, and the selected rate controls are removed from the DOM.

Is switching sample-accurate?

No. It keeps the nearest shared media time and clamps to the shortest measured output, but encoder delay can differ.

Does a higher bitrate improve a low-bitrate source?

No. It cannot restore discarded source information and may only make a larger re-encoded file.

Are the quality comparisons scientific?

No. The tool provides private personal listening and measured size/bitrate data, not controlled test results.

Can I download each version?

Yes. Each verified version has its own local object URL and download, cleared when you restart or leave.