Testing methodology

Reproducible results, not borrowed claims.

We test the processing path, verify the output with FFprobe, and publish measured examples generated from controlled synthetic audio. Private customer uploads are never reused as test material.

Fixtures

Pink-noise and sine components produce repeatable, non-private 44.1 kHz references in WAV, MP3, M4A, and MP4 containers.

Production engine

The live worker currently runs FFmpeg 5.1.9. Reference fixtures and the August 2026 benchmark pass were verified with FFmpeg 8.1.2.

Validation

FFprobe confirms duration, codec, sample rate, channels, bitrate, and readability after processing. End-to-end jobs also verify downloads.

Published benchmark set

Audio Compressor60-second, 44.1 kHz stereo WAV synthetic reference — 10.58 MBBalanced preset, MP3 at 160 kbps stereo1.20 MB

Compress Audio to Target Size180-second, 44.1 kHz stereo WAV synthetic reference — 31.75 MB5 MB target, calculated 228 kbps MP3 (mapped to an encoder-supported rate)5.04 MB decimal / 4.81 MiB

MP3 Compressor60-second MP3 at 320 kbps — 2.40 MBSpeech-oriented MP3 at 96 kbps0.72 MB

M4A to MP3 Converter60-second AAC-in-M4A reference at 128 kbps — 0.97 MBStandard MP3 at 192 kbps stereo1.44 MB

WAV to MP3 Converter60-second, 16-bit, 44.1 kHz stereo PCM WAV — 10.58 MBStandard MP3 at 192 kbps stereo1.44 MB

Audio Trimmer60-second MP3 at 320 kbps — 2.40 MBKeep 00:10–00:25 with MP3 stream copy0.60 MB for 15 seconds

MP4 to MP3 Converter30-second 640×360 MP4 with 128 kbps AAC audio — 0.51 MBExtract and encode MP3 at 192 kbps0.72 MB

Audio Format Converter60-second, 16-bit, 44.1 kHz stereo WAV — 10.58 MBConvert to Opus at 128 kbps0.69 MB

What the numbers do—and do not—prove

Synthetic references make settings reproducible, but they do not predict perceived quality for every voice, instrument, or recording. Codec behavior changes with spectral complexity, transients, channel relationships, source artifacts, and encoder version.

We publish sizes to explain the mechanics of duration and bitrate, not to promise that every input will produce the same byte count. Target-size results include container overhead and bitrate rounding, so strict upload workflows should keep a safety margin.

Review dates change only after the implementation, claims, citations, or benchmark results are materially checked. Corrections can be sent to contact@audio-totube.online.