Compress Audio Files Free: No Upload, Works Offline

Six formats go in and out: MP3, M4A/AAC, Opus, OGG, WAV, and FLAC. Mixing and matching is fine, so drop a folder of WAV, FLAC, and M4A files and send all of them to Opus in one batch.

There are eight presets, and each sets several things at once rather than just a bitrate. Podcast encodes to Opus at 32 kbps mono, normalizes loudness to −16 LUFS (the level most podcast hosts expect), and trims silence off the start and end. Podcast (MP3 compatible) does the same loudness work but outputs roughly V5-quality MP3, for players that still expect it. Voice Memo goes further on size: Opus at 24 kbps mono, resampled to 16 kHz, which is normally plenty for speech and happens to be the sample rate a lot of transcription tools want anyway. Music (share) and Music (small) favor listening quality: MP3 VBR around 190 kbps stereo, or Opus at 96 kbps for roughly half that size. Audiobook mirrors Podcast at a gentler −18 LUFS and collapses long internal pauses instead of only trimming the edges. Lossless archive re-packages WAV into FLAC at compression level 5, typically 50–60% of the original size with nothing thrown away. Custom hands you every control directly, and any preset can be overridden on a single queued file, so a batch doesn't have to be all-or-nothing.

Every job runs in one of two modes. Quality mode is what you'd expect from any converter: pick a format and a bitrate or VBR level, and the output size falls out of that choice. Target-size mode inverts it: type a number, like "fit under 100 MB," and the tool works backward from your file's duration to solve for the bitrate that lands there. It subtracts a small estimate for tag and cover-art overhead first, plus container framing overhead that varies by format (roughly 2% for Opus/OGG, closer to 0.5% for M4A), so the number you get is meant to land under your target, not just near it.

Bitrate control on compressed audio isn't perfectly precise, though. MP3 defaults to average-bitrate encoding and Opus to a constrained variable rate, so a first pass can occasionally land a percent or two over. When that happens, the tool re-encodes once more at a corrected bitrate rather than handing you an oversized file, which is why a target-size job sometimes takes two passes instead of one. There's also an "exact size" toggle for MP3 that switches to true constant bitrate for a harder guarantee, at a small cost to quality. And the math has a floor: push a file's target low enough and you'll get a warning instead of a silently ruined result: Opus stops being worth using below about 6 kbps, MP3 below 32, AAC below 24, with a suggestion to switch codec, drop to mono, or trim silence first.

WAV output is always 16-bit PCM, regardless of your source file's bit depth. If that matters to you, FLAC is the lossless option here instead.

AAC (M4A) doesn't have a working variable-bitrate mode in the engine this tool ships with, so picking a quality level for M4A maps to a fixed bitrate table instead. Below 96 kbps the format starts sounding rough enough that the tool suggests Opus instead.

Cover art carries over into MP3, M4A, and FLAC output, using whatever embedded-picture format each of those containers supports, but not into Opus or OGG: the art gets dropped, and you'll see a note about it rather than losing it silently.

Silence trimming only ever scans forward through a file, never backward. An earlier version trimmed trailing silence by reversing the audio, cutting, then reversing it back, but reversing audio means holding the whole decoded file in memory at once, which reliably crashed on anything longer than a few minutes. Scanning forward means loudness measurement and trimming can both stream through a multi-hour recording instead.