Lossless

What format is your digital music?

Most of mine are MP3s encoded at 256 kbps. That was a somewhat random choice made at the time that I ripped the bulk of my CDs a couple of years back.

Most … but not all, as I recently discovered when trying to move some of the music onto my Mac laptop. I found (more accurately: was reminded) that some of the albums were encoded in a lossless WMA format, which was pretty useless when it came to importing them into iTunes, so a converter was required [1].

On the surface the various digital music formats are a simple trade off between sound quality and file size.

At the top end (maximum quality, largest file size) are the so-called lossless encryption methods – as described in this Steven Levy article. The one I used was supplied with Windows Media Player and so created WMA files, but I notice that there is an alternative Apple Lossless format available within iTunes. I don’t expect either of these are commonly used.

All other alternative formats use some form of compression to create smaller files. MP3 is just one of those – although clearly the format of choice for the vast majority of people.

This is where it starts to get somewhat mysterious, even for people like myself who like to think they are somewhat technical. How much better is a 128 kbps MP3 vs. a 256 kbps MP3? Surely at some point you reach the limit of the CD you’re ripping from? How good does your stereo and speakers need to be in order to even hear the difference? Perhaps a reader will be able to enlighten us all on some of these things.

It does also beg the question as to why we’re compressing music at all.

There is a great story (a myth as it turns out [2], but let’s not let that get in the way) about how they chose the compression ratio when putting together the technical specifications for CDs so that the discs were as small as possible but still able to fit the entire length of Beethoven’s 9th Symphony. The result was a sampling frequency of 44.1 kHz.

That was 1981 and made sense because they were constrained by the capacity of the physical disc.

Likewise in 1994 the first computer I bought had a 200 MB hard drive, which was only just big enough to fit a few albums of MP3 files.

But today that constraint is well and truly gone. Disk space is now abundant. I can fit many more songs than I own on an iPod. Yet despite that we still compromise quality to keep file sizes down, which seems a bit odd.

Along the same lines …

Do you take photos at the highest resolution your digital camera allows?

The highlight of 2007 for me was the arrival of baby #2 in April. This provided a perfect excuse to splash out on a new digital camera, which came complete with 10 mega pixels (an embarrassing abundance, surely?)

However, after using it for a couple of days (and thankfully before the baby arrived) I was surprised to discover that by default the camera was set to a resolution much lower than the maximum it supported.

Why? I can only assume it was because the memory card that came supplied with the camera was so small that it would fill up almost immediately at the maximum resolution. But I had also splashed out on a bigger card, which had room for hundreds of images, so that constraint didn’t apply [3].

Thinking about this reminded me of my own baby photos, which are easily recognised as coming from the mid-1970s due to their square shape and rounded corners (not to mention the funny looking clothing that all the grown-ups are wearing!) By the time my younger brother and sisters came along the technology had moved on and so all of their photos are the more modern 6″x 4″ size.

I realised that the same will be true for my own kids too.

Technology moves on quickly, but we users are not always so quick to adjust to those things that are the constraints.

So, please, encode your music at the highest quality your software allows and make sure that your camera is set to the highest resolution [4].

You’ll be glad you did in a few years time.

Notes:

[1] If you run into similar problems I can recommend an excellent free utility called Switch which allows you to convert a WMA file into an MP3, albeit at the expense of the tags identifying the artist, album etc.

[2] As Kees Immink, one of the engineers who worked on the original CD specification explains, the choice of 44.1 kHz was really the only choice available to the engineers due to the recording equipment available at the time, and the size of the discs themselves was more of a management/marketing decision than an engineering one – they wanted them to be the same size as a cassette tape (as it turns out the engineers had the final say – the discs are 0.5 cm larger than cassette tapes at their widest point, providing CDs with significant additional surface area).

[3] This is another excellent example of software designers not thinking about the “Pit of Success”. Ideally the camera would have prompted me when I inserted the larger memory card for the first time to suggest I increase the resolution settings. Or, even better, the camera would come with a sufficiently large card in the first place!

[4] In case this suggestion seems to contradict my theme of less I should point out that I’m not suggesting more music or more photos just higher bits rates and resolutions.

As with anything, removing one constraint just highlights another.

In the past the constraint with photography was always the number of photos you could take – especially going back to the days of 24 frames per roll of film (younger readers, please ask your parents to explain this crazy concept). With a digital camera and lots of memory that’s no longer an issue. But, it occurs to me that the new constraint now is finding the great photos amongst the 1000s that you can so quickly and easily rattle off

(This, by the way, is a problem that technology might be able to solve – tools like Flickr’s Interestingness have started to do this automatically for a larger set of photos. Imagine if you could do something similar for the photos on an individual hard drive. There’s gold in them hills I reckon!)

If you take the time to delete the average photos you’ll enjoy the really good ones much more. Likewise with music – take some time to delete the tracks you don’t like from the albums you rip. Your ‘skip’ button will thank you for it.

So, perhaps the new constraint is time?

6 thoughts on “Lossless”

  1. I once found a CD ripping guide that was actually supposed to be a standard for illegal pirate groups to adhere to when uploading ripped copies to P2P networks. The recommendation in that guide was to rip at 256kbps variable bit rate (VBR). The VBR is the key as it allows the program to encode at a higher bit rate for the complex parts of the song, and can also encode at a much lower rate for the less complex parts of a song. This results in a similar file size as a standard 256kbps constant bit rate (CBR) song, but at a much higher quality. In fact, you’d be hard-pressed to hear the difference between a 256kbps VBR track and the original CD source.

    As for photos, I always take mine on the highest resolution possible as you can always scale it down, but you can’t scale them up. So if you ever want to print out some large copies, then you’ll wish you had taken the photos at the highest resolution.

  2. As a DJ, I always have my mp3 files at 320bps and sometimes, but rarely, at 192kpbs if I cannot help it.

    To be honest, the difference in sound quality between 256 to 320 is minimal (unless you’re an audiophile).

    The difference between 128 and 256/320 though, can be heard quite easily with decent speakers (say your standard Altec Lansing).

    The most apparent symptoms will be volume and loss of quality on the lower end (bass, mid-bass) of the track.

    If you cue a 128 tune, and have a 256 tune immediately right after, you’ll find that the volume of sound increases from the 128 to 256 tune, even if you didn’t manually change it.

    The clarity of the lows is also better on a 256 tune than it will be for 128, where often you’ll find that it sounds muffled.

    Of course, you won’t hear this difference if all your tracks are compressed to 128. But side by side with a higher bitrate, I think the difference is apparent.

    192s are also generally quite good though. They do sound a bit thinner than 256s still, and on a big system requires more work from the amps or gains to push through the sound as compared to a 256. But again, for home listening, 192s are great.

    The only possible way to keep full quality is to convert to WAV.

    Although if you’re ripping from CD, you will still lose some quality but it will be infinitely smaller.

    As a rule of thumb though, 320kbps files are about 1.5 to 2x the length of tune, in file size.

    So if the tune was 10 minutes, a 320 conversion will make it about 15MB or 20MB.

    128kpbs is about half the standard tune length, in file size.

    WAVs are a whole lot bigger from memory, and that is why most people don’t convert to WAV.

    If you’re really an audiophile however, don’t rip with WMP but maybe a sound editing or production programme like Soundforge, Ableton etc :)

  3. I agree with Stuart – VBR rocks. I use it for both MP3 and AAC, and any movies I create. Best solution for size and quality (I use either 256K or 320K VBR).

    That said, I still can’t fill up my 8GB Nano with stuff I wanna listen to when I’m out, so the size isn’t really an issue :)

  4. If I’ve only got a MP3 copy (i.e. its not a back up of a CD I have purchased): 320 preferred, but 256 is ok. I’ll listen to a mix at 192, but often you can hear a metallic, rasping sort of sound on 192 stuff. Then again I’ve got about a terabyte of storage on the Media Centre PC in the lounge so space isn’t an issue.

    For my backup of my CDs everything is about 192, but thats just because the ripping software and LAME mp3 encoder I have only does a max of 192.

    If its on a portable device I’ll probably go for 192 however, due to space constraints.

    “192s are also generally quite good though. They do sound a bit thinner than 256s still, and on a big system requires more work from the amps or gains to push through the sound as compared to a 256. But again, for home listening, 192s are great.”

    I think 192 is fine on a $200 but anything with decent subs, tweeters etc and it sounds horrible.

    “The most apparent symptoms will be volume and loss of quality on the lower end (bass, mid-bass) of the track.”

    Yes, especially considering what I listen to is bassheavy stuff, i.e. reggae, breaks, dnb and dubstep. For Dubstep you simply lose a lot of the track (which is in the sub bass frequencies) if its lower than 256. I’m sure other stuff which is more higher frequencies the loss in quality isn’t so apparent. I could be wrong however.

  5. Anything under 192 is questionable.
    If you have any electronic music such as Aphex Twin then I would not go under 256 if possible. Once you go under 256 the the highs sound less dynamic and hurt the ears more and the bass is more flattened.
    The only reason I don’t like VBR is because the players always show them being an odd bitrates (like 470kb or 678kb) so it makes it difficult when doing file management to classify the higher bitrate MP3s versus to low ones.

Comments are closed.