Lossless versus lossy audio

How to use this workflow

1. Identify the best source you have

   Keep the original recording or highest-quality available source. A WAV created from an MP3 is still derived from that MP3.

2. Minimize repeated lossy exports

   Keep an editable PCM or lossless intermediate where practical. Avoid creating a new MP3 after each individual cut, pitch shift and effect.

3. Create the final delivery version

   Choose the format needed by the destination, export once from the completed edit and compare it with your working version.

Lossless is a property of a coding step

FLAC and ALAC can encode samples without the perceptual losses of MP3, AAC or Vorbis. But if the samples came from an MP3 decoder, preserving them exactly does not reconstruct the recording that existed before the MP3 was made.

Processing also matters. Trimming, equalization, resampling and pitch shifting intentionally produce a different signal. Exporting that new signal in a lossless format preserves that stage’s result; it does not make the whole workflow unchanged from the original.

Compare size using stated assumptions

For sixty seconds of two-channel, 44.1 kHz, 16-bit PCM, sample data occupies about 10.58 MB in decimal units, before file overhead. At a constant 192 kbps, sixty seconds represents about 1.44 MB of encoded audio before overhead. These are calculations, not files downloaded or measured in a test.

A FLAC file’s size depends on the signal and encoder settings, so a fixed universal compression percentage would be misleading. Likewise, actual compressed file sizes include headers and metadata, and variable-bitrate formats need not match one simple bitrate calculation.

Evaluate the result rather than the label

Listen to the final file on the playback setup that matters to you. Compare exposed reverbs, sharp transients and dense high-frequency passages at similar levels. A bigger file can still contain the same damaged or distorted source.

This guide does not claim a blind listening study, a universal audible threshold or a guaranteed preference for one format. If you conduct your own comparison, use the same source and avoid level differences that bias the decision.

Choose an archive for the samples you actually have

For a finished PCM recording, FLAC can be a useful compressed lossless copy when your future player or editor supports it. WAV can be simpler for an editing handoff. Decide using a small representative file before converting an entire archive: storage savings depend on the signal, while compatibility depends on the receiving application.

In this converter, choosing FLAC with the same intended sample rate and bit depth keeps the settings decision separate from compression level. Levels 0–8 change the compression tradeoff, rather than offering eight tiers of audio fidelity. Changing sample rate or bit depth is a separate operation; “lossless codec” does not promise those choices leave the input samples untouched. Preserve the original when exact provenance matters.

Check an archive without confusing file bytes and audio samples

Export one file, open it in the destination, and inspect duration, channel count, sample rate and bit depth where that application exposes them. Listen at the beginning, a quiet passage and the ending. Copy it to the intended storage location and confirm that the stored copy opens before deleting any working copy.

A WAV and a FLAC containing equivalent audio have different file bytes, so their file checksums should not be expected to match. A checksum can verify that a copied FLAC file is unchanged during storage or transfer. Proving equality of decoded samples requires a separate sample comparison at matching format settings; this converter does not provide that proof. Keep required tags, artwork and alternate streams separately if the archive must preserve more than the chosen audio stream.

• Keep the source and record which settings were used.
• Verify the receiving application can decode the archive.
• Check a copied file, rather than only the immediate download.
• Retain the original for stream or metadata preservation requirements.

Keep one working master through several edits

Suppose you need to trim an introduction, change the key and add reverb. Saving MP3 after every edit means repeatedly decoding a lossy copy and encoding another lossy copy. A simpler working plan uses a PCM WAV between editing stages and creates the required MP3 only after the final edit. If your starting source is already MP3, decode it once for the working copy; this avoids additional intermediate MP3 encodes but cannot undo the original one.

The table counts new lossy encoding stages in two proposed workflows, not measured quality loss. The actual audibility depends on the source, settings and processing. A WAV intermediate still represents the edited signal at its selected sample rate and bit depth; it does not promise untouched source samples or unlimited numerical precision.

Swipe or scroll sideways to read all columns.

Start a revision from the appropriate saved stage

Keep filenames that describe the completed stage, such as song-trim.wav, song-trim-plus2st.wav and song-final-reverb.wav. If the recipient requests a lower-bitrate delivery, convert the final WAV again instead of converting the previous MP3. If the requested change is a different key, return to the pre-transposition working copy so the new shift is applied once.

Keep source and final delivery alongside the useful intermediates until the revision is accepted. Do not claim that a larger lossless file has recovered missing source detail. The advantage of this plan is a clear editing history and fewer lossy encoding stages, not a guarantee that successive pitch or effect operations remain acoustically identical.

Why a larger conversion can contain the same source limitations

Consider an invented one-minute stereo source delivered as a 192 kbps MP3. Its encoded audio is approximately 1.44 MB before overhead. Decoding it to 44.1 kHz, 16-bit stereo PCM WAV creates about 10.584 MB of sample data. That increase represents a different storage representation; it does not establish that more of the original performance was recovered. These are calculated sizes, not downloaded test files.

Before choosing a larger format, name the next task. For an editor requiring PCM, the WAV can be useful even though it retains the decoded MP3’s limitations. For a listener who already accepts MP3, a larger copy may add transfer cost without supplying a better source. For a restoration task, look for the original recording or a less processed master rather than assuming conversion can reconstruct removed detail.

Compare the same short passage at similar levels if you need to judge usability, and keep the best available source. Do not interpret a player’s newly reported bitrate or bit depth as the provenance of the recording. A FLAC made from the decoded MP3 can preserve those decoded samples under matching settings, but its compression size depends on the signal and it still has the MP3 in its history.

Swipe or scroll sideways to read all columns.

Choose FLAC compression without treating the level as sound quality

The FLAC compression-level control chooses encoding effort rather than a higher musical fidelity grade. For the same samples supplied to the lossless encoder, levels 0 and 8 are intended to decode to the same samples; the encoded files can have different sizes and different encoding times. Level 8 is not “eight times better,” and neither setting restores material removed by an earlier MP3 encoding. Sample rate and bit depth are separate decisions that can affect the samples supplied to the encoder.

For a useful comparison, add the same best source again and hold sample rate and bit depth constant. Create one FLAC at level 0 and one at level 8, then record their actual byte sizes and the elapsed times you observe on your device. Do not fill the results from an expected percentage: savings depend on the source and implementation. This is a proposed comparison, not a benchmark performed for this guide.

Choose a setting that meets the storage and turnaround needs, and reopen the accepted file in the receiving application. Keep the original source for metadata or alternate-stream requirements. If decoded-sample identity must be proved, compare decoded samples externally with matching properties; this converter offers no sample-equality report. A smaller FLAC file alone is not evidence of lost musical detail or proof that every other conversion setting stayed unchanged.

Swipe or scroll sideways to read all columns.

Common questions

Continue with your audio

• Compare the seven output formats →
• Make an editing intermediate →
• Make a delivery copy →
• Plan an ABX Audio Test and Read the Score →
• Schedule Short Audio Comparisons with Repeat Checks →