Methodology
How the audio conversion works
Every result is produced on your device. This page explains the engine, the settings and what the converter can and cannot do.
Last updated:
The conversion pipeline
The page loads FFmpeg compiled to WebAssembly from a public CDN the first time you convert. The browser reads your selected file with the File API, writes it into the WebAssembly filesystem, and runs an FFmpeg command with the output settings you choose.
When FFmpeg finishes, the browser reads the output bytes, creates a downloadable Blob and revokes it when you leave the page. AudioMorph does not receive your audio at any point.
Encoding settings
MP3 uses the LAME encoder, M4A uses AAC, OGG uses Vorbis, OPUS uses libopus, FLAC and WAV/AIFF use lossless PCM or FLAC encoding, and WMA uses the WMA v2 encoder. The bitrate selector applies only to lossy formats. Sample-rate and channel controls are passed directly to FFmpeg.
Optional volume normalization uses FFmpeg’s loudnorm filter with a target of -16 LUFS, -1.5 dBTP and an LRA of 11. It is intended to make speech or mixed clips more consistent, not to master music.
Quality and limitations
Converting a lossy file to another lossy format can reduce quality further. A higher bitrate cannot restore detail that the source already lost. For archiving, keep the original file and use a lossless format when storage allows.
Very large files, unusual codecs, DRM-protected audio and damaged files may fail. Browser memory, not a server limit, is the practical ceiling. The interface caps files at 200 MB, but lower limits may apply on mobile devices.
Privacy boundary
The audio engine is delivered by a third-party CDN, which necessarily sees the request for the engine (including an IP address). The engine request does not include your audio. Google AdSense, when enabled, is governed by the consent and cookie information on this site.