YouTube audio quality – How good does it get? (2022)

Original Article ↗ Hacker News Discussion ↗

Bitrate, buffering, and page bloat

  • Several comments argue YouTube optimizes more for startup latency than audio fidelity, assuming users hate buffering more than they value sound quality.
  • Others note that audio is tiny compared to video, so even doubling audio bitrate would add only milliseconds of buffering.
  • Some point out that page load bloat (JS, requests) dominates latency far more than media bitrate.

Adaptive streaming and codecs

  • People note YouTube already uses chunked, adaptive streaming (HLS/DASH-style), making bitrate ramp-up trivial.
  • Opus and AAC are used with variable bitrate by default. Example numbers: common Opus stream “251” is ~135 kbps in practice, not 251 kbps.
  • There’s debate over typical delivered video bitrates; “recommended for upload” vs “actual from YouTube” are very different, especially for low-motion content where audio can rival video in share of bits.

Hardware vs software for audio encoding

  • One side claims audio is such a small compute and bitrate cost that dedicated hardware (VCUs) isn’t needed.
  • Another questions this, citing YouTube’s heavy video acceleration; benchmark data is then presented showing modern CPUs can encode Opus hundreds of times faster than realtime, supporting the “audio is cheap” view.

Evaluating codecs and human hearing

  • Multiple commenters stress that frequency plots and spectrograms are misleading for lossy codecs; only double‑blind listening tests (ABX) are meaningful.
  • Discussion of psychoacoustic principles (equal loudness, auditory masking) and how codecs exploit them.
  • Acknowledgment that individuals differ: some find MP3 painful, others can’t distinguish it from live sound; “MP3 quality” is highly encoder/setting-dependent.

Playback speed and time‑stretching

  • Strong complaints about YouTube’s poor-quality time stretching for music practice at 0.75–1.1× speeds.
  • Disagreement over how “niche” this use case is; many say variable speed is heavily used, especially for courses and long-form content.
  • Some argue browsers, not YouTube, implement the time stretching; others counter that YouTube could drive better algorithms via JS/WASM or standards pressure.

YouTube Premium, Music, and “high quality” modes

  • Reports of an experimental “high-quality audio” toggle on mobile and existing “high” quality settings in YouTube Music (up to ~256 kbps AAC/Opus).
  • Some listeners say even ~135 kbps Opus should be transparent, yet YouTube’s output still sounds “smoothed” or lacking detail, suggesting other processing beyond simple bitrate limits.

Masters, remasters, and catalog instability

  • Commenters note that on YouTube Music, Spotify, etc., the underlying masters often differ from CDs: remasters, altered versions, or even different takes/singers.
  • People complain their streaming libraries silently change over time (tracks replaced, intros/outros altered, songs split to game per-track payouts).
  • Loss of original uncompressed masters and MP3-derived “lossless” files are reported as widespread problems in digital stores generally.

Technical details: sample rate, filters, and alignment

  • Some criticize including 44.1 kHz in the analysis because most consumer paths are effectively 48 kHz anyway, with resampling in software.
  • Others explain why oversampling and higher internal rates can be useful in ADC/DAC design but argue >48 kHz distribution is pointless.
  • A claim that YouTube now low‑passes many streams at ~16 kHz is made; a rebuttal says most people can’t hear above that and modern codecs already allocate few bits there, so savings aren’t as big as raw sampling-rate math suggests.
  • The article’s observation of a ~6.5 ms timing shift is called likely a bug or tool-chain artifact; multiple commenters say the overall comparison methodology is flawed for serious null testing.

Perceived quality and sentiment

  • Some insist YouTube audio is “obviously bad” and worse than old 320 kbps MP3s or P2P-era rips.
  • Others argue that modern Opus at ~128 kbps is generally transparent if encoded well, so complaints must stem from YouTube’s specific processing choices or expectations set by better references.
  • A few note seeing YouTube briefly serve higher-bitrate AAC (around 256 kbps) and then seemingly reverting to ~128 kbps, reinforcing the sense that YouTube is conservative on audio quality despite its massive resources.