Monday, February 06, 2012

Getting to Neil Young's audio quality

Musician Neil Young's interview with Walk Mossberg at the recent All things D: Dive Into Media conference last week is worth a look.

Young says that in the studio he's pretty happy with masters recorded at "24/192" which is presumably 24 bit at 192kHz sample rate (lossless compression). He bemoans the fact that most listeners currently get "just 5%" of the music data in the form of MP3 compressed audio.

I think that's overstating it a little, while the file size might be 5%, the genius of lossy compression is that it doesn't sound 95% worse. Despite apparently having been working with the late Steve Jobs on bringing better quality to portable players, Young never mentions better codecs such as AAC which Apple standardised on, Apple Lossless or the freely available FLAC.

My setup here has a number of external USB audio devices and to find out what their capabilities are you use the "Audio MIDI Setup" app found in the Utilities folder. My day to day headphone driver is a Corda "Move":

Like the Mac's internal audio, it's capable of 16/48, but interestingly it was defaulted to 16/44.1. This probably makes sense as pretty much all my music is from CDs which are sampled at 16/44.1kHz. I have some uncompressed tracks, you can see them in iTunes by getting info:

Again the sample rate is 44.1kHz and the sample size is 16 bit. I would assume that matching the output DAC to the sample rate of the recording would avoid resampling errors and extra CPU effort.

Would a move to 24/192 make a big difference?

Sampling at 44.1kHz means that the theoretical highest frequency that can be reproduced is 22kHz but a tone at that frequency would end up being a square wave that would be low passed filtered back to a sine wave - no doubt bearing little resemblance to the original waveform except in frequency. While I certainly can't hear beyond about 15kHz these days, I'm confident that others, particularly teenagers, can. Going to 192kHz seems a big jump. Perhaps 96kHz would be a good next step.

24 bits instead of 16 doesn't seem like a huge leap and given that storage and bandwidth go up by a third every few months I can't see any reason not to go to 24 bit.

Compressed bit rate

The other part of the equation is the bit rate that lossy compression must fit the audio in to.

Apple made the move to 16/44.1 AAC at 128kbps in the iTunes store without being pushed a few years ago and one of the benefits of their cloud music system is that they give you a better version of your tracks if required. (AAC is better than MP3 and a bit rate of 128kbps AAC is generally agreed to be equivalent to 196kbps MP3).

My guess is that further increasing the compressed bit rate would give a greater improvement to the perceived audio than increasing sampling or sample size bits. Of course no lossy compression is where we need to get in the end.

High resolution source audio

Increasing the sample rate or sample size beyond the original that most of us deal with every day - Audio CDs, will not improve the audio. Short of original recordings, the only other widely available source of high quality audio are "Super Audio CD" (SACD). These are recorded at a sampling rate of 2.8224 MHz  and encoded as direct stream digital which is hard to directly compare with 16/44.1 but given the claimed high frequency response limit of 100kHz, seems more than twice as good, and arguably good enough for Neil Young. (It would be interesting to quiz him about this).

I found some 24/96 audio and was able to play it with VLC (although my USB DAC will only sample up to 48kHz):

Turns out I have an audio device capable of playing this without loss. It's an Edirol UA-25EX. Once the driver was installed it can do what we want:

It sounds pretty good, particularly the cymbal sounds I think, but to be honest at my age it's probably lost on me.

Pay for quality option

Another interesting thread in the interview was that "piracy is the new radio". He's talking about customer learning about new music by first getting a copy from someone and then going on to buy the tracks.

We already pay a little extra for an HD version of a movie, (and I always do pay the extra personally), it makes sense to offer more options including very cheap versions for small screens.

In the case of music, where it is often being consumed in noisy environments or simply as low level background, it might be good to have very low priced - (or even free) to a low bit rate, possibly compressed and equalised to suit the listening environment. There would be the opportunity to "upgrade" to a high resolution version.

Young's comment about it taking 30 minutes to download a decent quality version won't prevent piracy of these versions as it's clear that consumers currently download HD movies without much sweat and speed will relentlessly increase.

What really makes the difference to audio listening quality?

Before all of these upgrades, the first thing to upgrade is speakers/headphones and amplifier. I recently purchased a pair of Sony MDR-7506 headphones and have been very happy with them. Day to day I use a little headphone amplifier and have built several Chu Moy amplifiers that really sound great.

Young is kind of right but doesn't realise that our existing hardware is already capable of increased quality in the form of lossless support and slightly higher bit rates. I'd like to see sliding price scales for better quality digital audio so that those who want it can get it.

Discussion on RN Breakfast

We had a panel discussion on RN breakfast this morning. Got a big listener response which was great. Here's co-panelist, Robbie Buck, standing by for air on the "brefast" show:

We had a bit of fun and played a vinyl record and then an MP3 for the listeners, many of them on AM radios. Lots of emails from people preferring the vinyl version.

iTunes store accepts higher resolution audio

I've just read in “Music as the Artist and Sound Engineer Intended,” that iTunes has asked publishers to submit high-resolution 24-bit/96kHz files, and not original CD masters to create the 256kbps files which are sold as iTunes Plus tracks.

Great stuff!

1 comment:

John Hancock said...

I find the whole audio quality debate extremely interesting. I am old enough to remember vinyl first hand and even had a 75/16 player in my younger days.

I was engaged in a vigorous debate a few years ago when someone posted a research paper from about (now) 20 years ago. The researchers used a high quality recording (cannot remember the actual source but it was high enough to make a difference) and channeled it through a 44.1k (CD quality) codec. The A test was the original and the B test was encoded then decoded to CD quality. They tested a large number of subjects of both sexes and from a wide range of backgrounds and experiences making sure ti include audiophiles who claimed a high degree of discrimination. What was most interesting was that the results indicated that no one could tell the difference between the sources. Some groups showed a slight but statistically insignificant bias towards the original but (as I recall) the audiophiles were split right down the middle.

I know that there are some rather slight, and quite esoteric limitations to CD quality but the bottom line was from the researchers was that by and large your regular sound source cannot be improved over CD quality or 44.1K.

Interestingly, if you think about it out sight is digital as is our hearing. We have some extremely sophisticated D-A circuitry in our heads.