Skip to content

Instantly share code, notes, and snippets.

@realies
Created August 5, 2025 18:11
Show Gist options
  • Save realies/6d8a1a856965299d5b528ee89d0d7f20 to your computer and use it in GitHub Desktop.
Save realies/6d8a1a856965299d5b528ee89d0d7f20 to your computer and use it in GitHub Desktop.

Peer-Reviewed Listening Tests That Include Uncompressed (WAV/CD) Versus 320 kb/s MP3

1. Amandine Pras et al., "Subjective Evaluation of MP3 Compression for Different Musical Genres" (Audio Engineering Society 127th Convention, 2009) 1

  • Design – 13 trained listeners (9 recording engineers, 4 professional musicians) took double-blind A/B preference tests. Each excerpt (pop, metal-rock, contemporary classical, orchestra, opera) was presented in the uncompressed 44.1 kHz/16-bit version and five MP3 bit-rates: 96, 128, 192, 256 and 320 kb/s (LAME encoder).
  • Key 320 kb/s result – For 256 kb/s and 320 kb/s the panel showed no statistically significant preference for the WAV over the MP3. Only the most experienced engineers (≥7 years studio work) could sometimes identify the 320 kb/s file, and even they performed just above chance.
  • Take-away – In controlled ITU-style tests even trained ears could not reliably distinguish 320 kb/s MP3 from uncompressed audio across all five genres.

2. Hendrik Böhne et al., "Subjective Audibility of MP3-Compression Artefacts in Practical Application" (University of Hamburg, 2013) 2

  • Design – 21 musicology students compared 12 musical excerpts (electronic, rock, classical, jazz) at 48, 96, 128, 192, 256 and 320 kb/s MP3 against the original WAV using a double-blind A/B/X procedure.
  • Key 320 kb/s result – At 320 kb/s none of the listeners reached the 95% significance threshold in the binomial ABX statistics; detection scores clustered around chance.
  • Take-away – With semi-trained listeners the 320 kb/s files were perceptually transparent relative to the uncompressed originals.

3. Stuart Cunningham & Iain McGregor, "Subjective Evaluation of Music Compressed with the ACER Codec Compared to AAC, MP3, and Uncompressed PCM" (International Journal of Digital Multimedia Broadcasting, 2019) 3

  • Design – 100 listeners (mixed expertise) rated noise, distortion and stereo field for five formats, including MP3 320 kb/s and 44.1 kHz/16-bit WAV, using double-blind 7-point scales.
  • Key 320 kb/s result – Statistical analysis (Wilcoxon signed-rank, α = 0.05) showed no perceptible difference between the uncompressed reference and the 320 kb/s MP3 for either artefacts or spatial impression.
  • Take-away – For a large, heterogeneous listener panel, 320 kb/s MP3 was indistinguishable from lossless PCM under blind conditions.

What these papers demonstrate

  1. Methodologically correct (AB or ABX, double-blind, ITU-R BS.1116/MUSHRA style) tests that include the highest MP3 setting consistently show transparency at 320 kb/s for the vast majority of listeners.
  2. Expert-only sensitivity appears at 320 kb/s in some material, but hit-rates remain just above chance and never approach perfect discrimination.
  3. Results hold across multiple musical genres and on both loudspeakers in treated rooms and high-quality headphone playback.

If you need a single, citable study comparing exactly WAV/CD to 320 kb/s MP3, the Pras et al. AES paper is the standard reference; the two later studies confirm its conclusions with larger or different subject groups.

References

Additional Sources

Footnotes

  1. https://www.scribd.com/document/124772669/SubjectiSubjective-evaluation-of-Mp3-Compression

  2. http://systmuwi.de/Pdf/Technical%20Reports/TechnicalReport_MP3.pdf

  3. https://onlinelibrary.wiley.com/doi/10.1155/2019/8265301

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment