Let’s put an end to the debates around sample rates in digital recording! We’ll BUST 12 myths that are still flying around.
By Craig Anderton
Here’s our first myth: Now that digital audio is several decades old, questions about sample rates have all been resolved.
Busted! Some people have opposing viewpoints about almost every aspect of digital audio and are certain they are correct. So, we may not end up totally busting a myth because someone, somewhere, will continue to believe it—and may even be right! But we’ll do our very best to at least dent these myths as much as possible.
1. High sample rates are important because even if we can’t hear certain high-frequency sounds, we can feel them.
For the sake of argument, suppose we can perceive frequencies above 20 kHz. If even one piece of gear in the chain from mic to speaker can’t reproduce these frequencies (and almost all consumer equipment cannot), then the point is moot. And if the gear can reproduce those high frequencies, you’re asking your speaker to reproduce audio you can’t actually hear, which may lead to intermodulation or other problems that degrade the sound.
Bottom line: If live music creates frequencies that recorded music can’t reproduce…listen to live music.
2. “I can’t hear any difference between material recorded at 96 kHz and 44.1 kHz.” And its twin brother, “recording at 96 kHz sounds repeatably and obviously better.”
They’re both myths—and both true.
For sounds entering a computer through an audio interface, I’ve never met anyone who can differentiate reliably between music recorded at 96 kHz or 44.1 kHz on the same system (assuming it has quality components). But sounds generated inside the computer by plugins like virtual instruments and amp sims can generate harmonics whose frequency exceeds the sample rate. This causes aliasing (a non-harmonic, audible distortion) because the audio can’t be sampled accurately. For these sounds, recording at 96 kHz sounds better—listen to the audio example.
- Example 1a – Synth 44.1 kHz
- Example 1b – Synth 96 kHz
The only difference between these two files is that they were recorded at two different sample rates. Recording the synth plugin at 44.1k has trebly artifacts while recording it at 96k sounds cleaner.
3. Because aliasing can be an issue, if you record at higher sample rates, your problems are solved.
Aside from practical considerations, like higher sample rate material requiring more storage and limiting the number of tracks you can stream, your converters may not perform as well at higher sample rates. Also, not all plugins work at all sample rates. (Waves has posted a helpful document that lists maximum sample rates for their plugins).
Furthermore, a higher sample rate may introduce artifacts that wouldn’t happen at a lower sample rate. Finally, modern plugins often include internal oversampling as they do their processing—the plugins are essentially running at a higher sample rate than your project, so they don’t produce aliasing. Then, they band-limit the resulting audio and downsample it back to the project sample rate. This allows the benefits of recording at a higher sample rate in projects with lower sample rates. For example, you won’t hear any difference running the Element 2.0 synth at 44.1 or 96 kHz.
4. Using plugins that oversample means you don’t have to record at higher sample rates.
Some argue that a plugin’s real-time, upsampling/downsampling process negates the “cleaner” approach of simply recording at a higher sample rate, which doesn’t need sample rate conversion. Granted, offline sample rate conversion can theoretically be more accurate than real-time algorithms. But can anyone hear a difference?
Reality check: If you record a great song, no one will care what sample-rate conversion algorithm the song used.
5. If you record at 96 kHz to minimize aliasing, you’ll lose any benefits because you’ll likely need to downsample it eventually to 44.1 or 48 kHz anyway.
It may seem counter-intuitive that rendering a 96 kHz file to 44.1 kHz doesn’t degrade the sound quality. However, as soon as you render the 96 kHz file as audio, then it’s audio, and audio has no problem playing back at 44.1 or 48 kHz, as you’ll hear from the audio examples. Aliasing is a problem only when frequencies above the sample rate can interact with the clock.
- Example 2a – Synth 1x sampling
- Example 2b – Synth 2x sampling
- Example 2c – Synth 4x sampling
These three files show how much oversampling can alter the sound of a harmonically rich synth waveform. The sampling rates are 44.1 kHz, 88.2 kHz, and 176.4 kHz, but even when downsampled to 44.1 kHz, the sonic differences remain the same.
6. The 44.1 kHz sample rate, with 16-bit resolution, was chosen for CDs because Sony executive Norio Ohga wanted Beethoven’s 9th Symphony to be able to fit on one disc.
Although Philips states this in their history of the CD (and they should know), others aren’t sure. However, what we do know is that early CDs were mastered on Sony’s 3/4″ U-Matic recorders. 44.1 kHz related to the scanning rate and worked with both NTSC and PAL formats.
7. Nyquist’s theorem is correct and has been proven to be correct for over half a century. So, any talk of higher sample rates is ridiculous.
Nyquist’s theorem is correct, but implementing that theorem requires filters that can eliminate all audio above the Nyquist frequency (i.e., half the clock sample rate). Because designing a totally transparent brick-wall filter isn’t easy, the simplest way to get around filter limitations is to use a higher sample rate. Then, the filter slope needn’t be so drastic. Some people believe that any preference for higher sample rates isn’t because of the sample rate itself but from being able to use gentler filters.
8. Video professionals use 48 kHz because it sounds better.
Theoretically, 48 kHz should sound better than 44.1 kHz. But using 48 kHz for video also makes calculations much easier. With a 24 frames-per-second rate, at 44.1 kHz, a frame equals 1,837.5 samples—but at 48 kHz, it’s a calculator-friendly 2,000 samples.
9. The Nyquist theorem applies only to digital technology.
Analog systems have the same constraints. For example, the bucket-brigade devices in analog delay signal processors (remember the Reticon SAD1024 and Matsushita MN3005 integrated circuits?) were subject to the same rules regarding sampling rate, aliasing, and so on.
10. Because a higher sample rate takes more samples, the waveform will be smoother.
Nyquist was right: Sampling a 20 kHz signal at 44.1 kHz provides sufficient data to reproduce the signal accurately. But sampling at 96 kHz also provides sufficient data to reproduce the signal accurately. The issue isn’t reproducing the waveform but reconstructing it into analog. Sampling sub-systems include multiple other elements, like filters and converters. I think most engineers would rather listen to 48 kHz audio through an excellent D/A converter than 96 kHz audio through a cheap D/A converter.
11. Ultimately, it really doesn’t matter what sample rate I use when recording because my DAW can convert the final mix to whatever sample rate I want.
While true, the quality of conversion algorithms varies. Visit the site SRC Comparisons if you’re curious as to how different DAWs perform – and expect some surprises. Also, note that some DAWs improve over time. For example, Ableton Live 7 had poor sample rate conversion, but Ableton Live 10 has excellent sample rate conversion.
12. Bonus trick question to win bar bets!
When you listen to a stereo file with a 96,000 Hz sample rate, how many unique samples do you hear in one second?
Most people will say 96,000—but the correct answer is 192,000 because stereo has two channels.
Musician/author Craig Anderton is an internationally recognized authority on music and technology. He has played on, produced, or mastered over 20 major label recordings and hundreds of tracks, authored 45 books, toured extensively during the 60s, played Carnegie Hall, worked as a studio musician in the 70s, written over a thousand articles, lectured on technology and the arts (in 10 countries, 38 U.S. states, and three languages), and done sound design and consulting work for numerous music industry companies. He is the current President of the MIDI Association. www.craiganderton.org.
Check out more audio myth busting. 12 Mastering myths BUSTED!
Want to get more tips straight to your inbox? Subscribe to our newsletter here.
Leave a Reply