Lossy encoders rebuild the waveform from a frequency-domain approximation, so the decoded signal can peak above the original — levels that show no overs on the PCM master can still clip after encoding. That is why Apple’s Digital Masters workflow (afclip / AURoundTripAAC) auditions every master through a 256 kbps AAC round-trip before release, and why AES TD1008 recommends keeping the true peak at or below -1 dBTP at the codec input: overshoot grows as the bitrate drops (an AAC 128 encode can add roughly +0.4 dB; lower bitrates more). This tool runs that round-trip for the codecs your listeners actually get: MP3 (LAME) at 320 and 128 kbps, AAC at 256 kbps (Apple Music tier) and 128 kbps (web-player tier), Opus at 160 kbps (YouTube Music high) and 96 kbps. Each selected codec encodes your master, the encode is decoded back to PCM, and both sides are measured with the BS.1770-4 oversampled meter: integrated loudness shift, true-peak overshoot (decoded peak minus master peak), headroom against 0 dBFS and against the -1 dBTP codec-input recommendation, plus file size. Each preview is graded: “clips after decode” (decoded true peak above 0 dBTP), “tight headroom” (above -1 dBTP) or “safe”. The encoded preview files and a comparison.json are packaged as a ZIP so you can A/B the actual artefacts in your DAW or player. Encoders differ per platform (Apple’s own AAC encoder is not the one used here), so treat the numbers as a revealing preview, not the exact stream.