Lossy vs. Lossless: The Two Philosophies of Compression

Why is FLAC half the size of WAV with zero quality loss, while converting MP3 to FLAC makes the file bigger without improving the sound?

A four-minute CD-quality song is roughly 40 MB in WAV format, about 20 MB in FLAC, and about 9 MB as a 320kbps MP3. The difference between these three goes far beyond "file size": convert the 40 MB WAV to 20 MB FLAC, and the quality is unchanged; convert the 20 MB FLAC to 9 MB MP3, and the quality degrades; convert that 9 MB MP3 back to a 40 MB WAV, and the quality does not recover. This one-way degradation is the single most important thing to understand about lossy compression.

The Fundamental Difference

Lossless compression reduces file size without discarding any data. The decompressed output is bit-for-bit identical to the original. It is a strictly reversible transform.

Lossy compression discards information that the human eye or ear is not sensitive to, achieving far higher compression ratios than lossless can reach. The process is irreversible—the discarded data is gone forever.

An everyday analogy:

• Lossless: you take a 300-page book and reformat it into a tighter layout—removing extra whitespace, merging repetitive elements—cutting the page count in half, but every word is still there, letter-perfect;
• Lossy: you translate the original into a condensed summary. The word count drops drastically, the gist survives, but the original's phrasing, nuance, and stylistic detail are gone for good.

An Information-Theory View: Redundancy vs. Essential Information

Why can audio, video, and images be compressed at all? Because files contain abundant redundant information:

• Statistical redundancy: data that appears frequently can be represented with shorter codes (Huffman coding, arithmetic coding)—this is the foundation of lossless compression, and all compression methods (including lossy ones) use it on the input signal first.
• Perceptual redundancy: the human eye is less sensitive to fine high-frequency detail, and the human ear is less sensitive to certain frequencies—this is where lossy compression "cheats," discarding only information that our senses don't much care about.
• Spatial/temporal redundancy: adjacent pixels or frames look similar—this is the domain of video compression, which can save space by recording only the differences, and lossy methods can take a further step by discarding small differences.

The theoretical limit of lossless compression is set by entropy—the irreducible information content of a signal. In practice, lossless compression ratios hover between 2:1 and 5:1, which isn't enough for photos, music, or video—hence the need for lossy compression.

Lossless and Lossy Formats at a Glance

Audio

Video

Images

Generational Loss: Every Re-Encoding Is a Subtraction

This is the most commonly overlooked cost of lossy compression:

Original file → encoded to MP3 320kbps → first lossy pass discards information
             → decoded back to WAV → encoded again to MP3 192kbps → second lossy pass on already-damaged information

Every "decode → re-encode" cycle in lossy compression introduces fresh loss—this is generational loss. The reason: each time the encoder sees a signal that has already lost some detail, it still follows its own parameters to encode that signal as efficiently as possible, and the old errors compound with new ones.

Practical patterns of generational loss:

• Same format, same bitrate (MP3 320kbps → decode → MP3 320kbps): the first pass dropped some detail, and the second pass may drop a little more at the same bitrate. At generous bitrates the effect is usually small.
• Same format, lower bitrate (MP3 320kbps → decode → MP3 128kbps): the loss accumulates noticeably; one downshift can be audible.
• Cross-format (MP3 → WAV → FLAC → AAC): the first two steps preserve the decoded MP3 signal losslessly (WAV and FLAC), but the AAC encode operates on a signal that has already gone through one lossy pass.

The golden rule: always begin with the original file (raw / WAV / uncompressed video) when creating a lossy encode. If all you have is a lossy source, treat it as "the final version"—don't try to regain quality by upconverting or re-encoding repeatedly.

Why "Converting Lossy to Lossless" Is Pointless

"Converting a 128kbps MP3 to FLAC made the file bigger—does that mean the sound quality improved?"

No. A lossless format only guarantees that decompressing it yields the same file you fed in—and that input is already a 128kbps MP3 that has discarded most of the original detail. Converting to FLAC simply wraps the decoded MP3 signal in a lossless package. The file grows, but the contents are still "the compressed signal"; the discarded information never returns.

Memorize these equivalent scenarios:

128kbps MP3 (roughly 80-90% of information already lost) → FLAC   (saves the MP3's quality, adds nothing)
Original WAV → FLAC                                                  (saves every bit of the original — perfect)
Original WAV → 128kbps MP3 → FLAC                                    (same quality as the 128kbps MP3; a lossless box holds lossy goods)

Lossy vs. Lossless: A Decision Tree

Common Misconceptions

1. "Lossless means uncompressed." Lossless compression can still shrink files significantly while preserving every bit—41 MB WAV → 20 MB FLAC is lossless compression, not "no compression at all."
2. "Converting MP3 to WAV will improve the sound." MP3 → WAV just puts an already-damaged signal into a larger box. What was lost is lost.
3. "FLAC always means CD quality, regardless of its source." FLAC inherits the quality of whatever went in. A 128kbps MP3 converted to FLAC sounds like a 128kbps MP3.
4. "Lossless is the only ethical format; lossy damages your ears." In practice, 256kbps AAC or 320kbps MP3 is indistinguishable from lossless in most listening conditions. Lossy is perfectly fine for everyday listening—no guilt required.
5. "Repeated re-compression doesn't matter—I'm just sending a file." Generational loss accumulates. For important footage, go from the original file to a single lossy encode; avoid chaining encodes.

Practical Tips

• Keep a master mindset: always retain one lossless original of any important audio or video. Your pipeline should be "original file → lossless archive → lossy distribution," not "original file → lossy encode → delete the original."
• Check the encode history: when handed a file by someone else, check whether it has already been through lossy compression. A PSD or TIFF is probably original; a JPEG is all but guaranteed to be lossy.
• For daily listening, lossy is enough: 256kbps AAC or 320kbps MP3 is indistinguishable from lossless on the vast majority of playback gear. The storage you save can go toward curating more music.
• Video is inherently lossy in distribution: there is no practical lossless video distribution (except on internal networks or review workflows). Use a high-bitrate lossy encode (CRF 18-20 or an adequately high VBR) for a "virtually lossless" viewing experience.
• Archive audio as FLAC: FLAC is open, widely compatible, supports metadata tagging and error-checking, and is the standard format for music archiving.

Further Reading

• Video Compression: Finding the Balance Between Quality and File Size — the specific parameters of lossy video compression
• Containers vs. Codecs: The Packaging and the Contents of a Video File — codec type determines whether compression is lossy or lossless
• Format Conversion: Making Files Flow Freely Between Platforms — practical format choices during conversion

This site's format conversion tools handle both lossy and lossless formats—and after reading this article, you'll know where a FLAC comes from, how hard a video encoder is working to compress your MP4, and which path to pick for each scenario.

Related Tools

• Video Converter
• Audio Converter