WAV vs. AIFF vs.
ALAC and Other Computer Shenanigans
Most computer users say "Lossless is Lossless" and "bits are bits" thus WAV, AIFF, FLAC and ALAC (Apple Lossless) should all sound the same, however do they really? Even so, some computer users hear differences between them. I would like to relate my experiences and put forth possible reasons for the differences I hear.
Unlike uncompressed music files, lossless compressed music files have to be uncompressed in real time (meaning while the music is playing, and in my opinion, this process might effect the sound). Losslessly compressed music files are bit by bit identical and can be converted back to an uncompressed format (WAV, AIFF) and sound like the original music file again with no losses.
I didn't compare FLAC as I own a Mac Mini and it's not supported by iTunes, however its codec works similar to ALAC (Apple Lossless).
I compared audio file formats using my Reference Recordings HRx 24-bit 176.4 kHz data DVDs and MA Recordings 24-bit 88.2, 96 and 176.4 kHz data DVDs directly imported into my Mac Mini in their native WAV format. I used XLD to convert them to AIFF and ALAC as I find XLD sounds better in file conversion that does iTunes.
WAV Audio Files Explained
WAV (Waveform Audio File Format) is a Microsoft and IBM standard for storing audio data. It's an application of the Resource Interchange File Format (RIFF) bitstream format method for storing data in "chunks", and thus is also close to the AIFF format used on Macintosh computers. It's the main format used on Windows systems for uncompressed audio. The usual bitstream encoding is the LPCM (Linear Pulse-Code Modulation) format.
WAV is also the format chosen for Reference Recordings HRx's, MA Recordings, and other Data-DVDs. WAV is the most popular audio file format used by recording companies for making new master recordings. WAV is also used by studios and mastering labs, all phases of modern recordings are done on the computer. Some recording companies use AIFF, or DSD's computer codec DFF instead. So the DVDs derived from these master music files are perhaps the closest we will ever get to the actual master.
AIFF (Audio Interchange File Format) is uncompressed PCM (Pulse-Code Modulation). There is also a compressed variant of AIFF known as AIFF-C or AIFC, with various defined compression codecs.
With the development of the Mac OS X operating system, Apple created a new type of AIFF which is, in effect, an alternative little-endian byte order format.
Because the AIFF architecture has no provision for alternative byte order, Apple used the existing AIFF-C compression architecture, and created a "pseudo-compressed" codec called sowt ('twos' spelled backwards). The only difference between a standard AIFF file and an AIFF-C/sowt file is the byte order; there is no compression involved at all.
Apple uses this new little-endian AIFF type as its standard on Mac OS X. When a file is imported to or exported from iTunes in AIFF format, it is actually AIFF-C/sowt that is being used. Apple refers to the files simply as AIFF, and uses the .aiff extension.
For the vast majority of users this technical situation is completely unnoticeable and irrelevant. The sound quality of standard AIFF and AIFF-C/sowt are identical, and the data can be converted back and forth without a loss. Users of older audio applications however, may find that an AIFF-C/sowt file either will not play, prompt the user to convert the format on opening, or will play as static.
ALAC (Apple Lossless Audio Codec) is an audio codec developed by Apple Inc. for lossless data compression of digital music. After initially being proprietary for many years, in late 2011 Apple open-sourced and made it royalty-free so we will be seeing its increase as a download option in the near future.
Lossless data compression is a class of data compression algorithms that allows the exact original data to be reconstructed from the compressed data. The term lossless is in contrast to lossy data compression, which only allows an approximation of the original data to be reconstructed, in exchange for better compression rates.
Apple Lossless audio files are 40-50% smaller, and I think Apple Lossless sounds great if not compared directly to AIFF or WAV.
Endianness Could Explain Differences in Sound Quality
Some have suggested the sonic differences between WAV and AIFF has to do with their "endianness". Both WAV and AIFF music audio files are compatible with Windows, Macintosh, and Linux operating systems. However Mac PowerPC's are big-endian, whereas the newer Intel Macs made since 2006 are little-endian and Windows PCs are little-endian.
If you have a different operating system or architectures you can check your endianness here: http://en.wikipedia.org/wiki/Endianness#Endianness_and_operating_systems_on_architectures
The term 'endian' refers to the ordering of individually addressable sub-components within the representation of a larger data item as stored in external memory. Each sub-component in the representation has a unique degree of significance, like the place value of digits in a decimal number. Endianness is a difference in data representation at the hardware level and may or may not be transparent at higher levels, depending on factors such as the type of high level language used.
The usual contrast is whether the most significant or least significant byte is ordered first. A big-endian music file stores the most significant byte first and a little-endian music file stores the least significant byte first.
As explained above in the AIFF section there are two kinds of AIFF files: little-endian or big-endian depending on what computer they were created on. In iTunes both types of AIFF audio files say AIFF, perhaps so the novice user is not confused. Instead go to "Finder", click on "Music", open the "iTunes" folder, select "iTunes media", then "Music" and then select an album folder you know is in the AIFF format. You will notice the songs have an .aiff extension after the name, it will never show the "C" here either, we have just a couple of more steps. Right click on a song, select "Get Info", this will open up an info box and under "kind" if it says "AIFF audio" it is big-endian, if it says "AIFF-C audio" it is small-endian.
WAV files are almost always little-endian. Big-endian WAV files are extremely rare so you likely won't ever run into one.
Since 'endianness' has to do with the order of the bytes (determined by the computer it was created on) it will always playback in the same order without regard to your computer being big-endian or little-endian.
To me, WAV sounds the best on my Mac Mini, which seems illogical since WAV is a Microsoft/IBM format. Endianness does not explain the differences I hear as my Mac Mini is an Intel Mac, thus when I create an AIFF audio file it is AIFF-C/sowt and little-endian, as are WAV files. However, I accept what I hear, even when the results are totally unexpected. It's possible that file conversion is not a perfect science and what I am hearing is a result of the differences in how the file is being played.
WAV has an ease of presentation and ambiance that sounds very lifelike; AIFF is very close. Both WAV and AIFF have a larger soundstage and a lifelike attack of percussion instruments that was somewhat subdued with ALAC (Apple Lossless)—although it sounded excellent otherwise. I would not expect one not to know what was missing, unless if it was played right after a WAV or AIFF music file.
There is an application for playing FLAC audio files in iTunes, however I have read that it has issues so I have never tried it. Instead, I use XLD to play FLAC files (http://www.macupdate.com/app/mac/23430/x-lossless-decoder). Up until very recently, I had been converting my FLAC files to ALAC (Apple Lossless), but I discovered that I liked the sound of uncompressed WAV and AIFF better. I think WAV has a slight sonic edge over AIFF, but it doesn't support album artwork so I am currently using AIFF.
ALAC, WAV, and AIFF support music audio files up to 32-bit 192kHz in iTunes and up to 384kHz with other software players. However, this does not mean that one will always realize all the resolution in a music audio file as it depends on the computer sound card, core audio, or the maximum resolution of an external DAC. For example, my 192kHz music audio files are downsampled to 96kHz.
If you decide to convert 24-bit ALAC files to WAV or AIFF, don't use iTunes as it will convert them to 16-bit, even though it retains the correct sampling frequency. Instead use XLD as it will retain the original 24-bit rate and sampling frequency.
My MAC Mini's operating memory is 1GB, so it's possible the act of unpacking losslessly compressed music audio files in real time accounts for the difference in sound quality between lossless and uncompressed files. Many recommend a minimum of 4GB of operating memory and at least 10% of free memory for playing music.
I recommend converting your favorite high resolution music audio files to both uncompressed (WAV, AIFF) and losslessly compressed (ALAC, FLAC) formats and chose the one that sounds the best. Computer memory is getting very inexpensive so we are way past the time of needing to keep our music files small.
24-Bit 48kHz Audio Files for My iPod
I recently converted some of my 24-bit 96kHz music files to 24-bit 48kHz in ALAC (Apple Lossless) to play on my iPod, and noticed that the sound quality was closer than I expected. However the 24-bit 96kHz versions clearly had more ambiance, somewhat smoother string tone, and felt more lifelike. My feeling is that the bit rate is more important in PCM than the sampling frequency, and while 16-bit is just not enough, 24-bit sounds very analog-like in my system... and on my iPod.
Of course I tried AIFF and WAV versions of my 24-bit 48kHz music audio files and realized similar sonic improvements as heard from my 24-bit 88.2kHz and higher music audio files. However I discovered an issue with my iPod Shuffle. While it will play ALAC (Apple Lossless) up to 24-bit 48kHz, AIFF and WAV files are restricted to 16-bit 44.1kHz.
Beware Fake High Resolution Downloads
To be honest, I never cared much for CD or 16-bit digital. What impresses me about high resolution digital is how the best sounds almost analog-like and nothing like 'digital' from a CD. However, one has to be careful as a lot of what is offered as high resolution is fake (just upsampled 16/44.1k)—one can use tools like Audacity to confirm this. I have learned that a lot of major labels are guilty of this fakery, so I mostly stick to audiophile labels.
Before purchasing high resolution music check out "Music Analysis - Objective & Subjective" at Computer Audiophile: http://www.computeraudiophile.com/f14-music-analysis-objective-and-subjective/
The Music Analysis forum was created to help computer audiophiles make educated purchases of physical formats and downloaded music. This area is for objective and subjective music analysis. Each thread is limited to discussing a single album. Objective analysis can include Spectrograms / FFT graphs, Waveform screenshots, and Dynamic Range information. Many phony high resolution downloads have been exposed resulting in them either being redone by the providers, or removed from their offerings. Computer Audiophile is the number one forum for keeping high resolution music providers honest.
My favorite high resolution digital recordings are from Reference Recordings, Telarc, and Chesky. I don't care for most major label "commercial" recordings except for the better audiophile re-masters which seem to squeeze greater realism from the recording, along with more ambiance and greater frequency extension... in both directions.
Still even after accepting digital preservation of music, my favorite music audio files on my computer are from audiophile LPs.
Nyquist Sampling Theorem
In essence, the theorem shows that a band limited analog signal can be perfectly reconstructed from an infinite sequence of samples; if the sampling rate exceeds twice the highest frequency of the original signal.
I cringe every time someone brings up Nyquist to support low resolution digital, as we need to move forward and not backward. Thus "perfect" is not the word I would use to describe the Nyquist theorem. I believe the interpretations of how the theorem works are flawed.
There are people who use the Nyquist theorem in an attempt to prove that 16-bit 44.1kHz music files are perfect and cannot be improved. These people say the Nyquist theorem proves any sample rate can perfectly reproduce the waveform beneath its Nyquist Frequency (half the frequency of the sample rate). I don't buy this since the bass is warmer and fuller with more impact in 24-bit than in 16-bit PCM. As the sampling frequency increases the bass gets more realistic... and I know bass instruments have no overtones in the ultrasonic range.
With increased sampling frequencies we not only get increased frequency response, but faster transient response. Meaning the impact of percussion instruments is more intense. Also the mid frequencies are smoother.
Quite revealing are musical programs limited to 20k by analog band-limited devices such as microphones or mixers. When doing a plot spectrum with these types of files you will notice they are not 'brickwall', but instead have a smooth natural roll-off; sort of like sliding down a hill instead of jumping off a cliff. These music files sound even more realistic from the deepest bass to the highest treble when sampled at even higher frequencies. It is not just the frequency range but also how often a sample is taken; 96,000 samples per second versus 44,100 samples per second provides much smoother steps from one sample to the next.
I prefer recordings with real ultrasonic response; naturally frequency limited recordings with a smooth roll-off sound better at higher (faster) sample rates.
Internal/External DAC Setup Recommendations
When comparing the internal DAC of a computer to an external DAC using the combination analog out/digital optical out port, remember your computer's output volume is disabled when using the "digital outs" to an external DAC. So check your computer's output volume when unplugging the digital cable and re-plugging the analog cable. If your computer is like my Mac Mini, it lowers the volume from 100% to 50% which drastically decreases the resolution. I have discovered my computer's output volume MUST be set at 100% for good sound quality; anything less sounds bad. Thus volume in Core Audio, System Preferences, and in iTunes must be set at maximum level (100%) and never lowered as it will drastically reduce resolution. This is because one loses resolution when decreasing volume in the digital domain, so instead control the volume with your preamp.
DSD Guide is a new website operated by Blue Coast Records and Cookie Marenco who are big supporters of DSD. They offer a few free DSD downloads to check out the sound quality of their new DSD recordings. http://dsd-guide.com/free-downloads DSD recording and re-mastering have been around since 1998 and is the digital format used for SACDs. DSD for computers is relatively new, I hope to try a DSD DAC towards the end of this year or early next year.
"Conversation with Cookie Marenco and Michael Bishop about DSD at T.H.E. Show 2012" is on YouTube http://www.youtube.com/watch?v=HFRG6IMLRV4
Don't forget, there are still plenty of free sample 24-bit PCM downloads whose selections change all the time. I noticed some of my links to them in previous articles no longer work, so I suggest doing a web search using the words: 24-bit downloads.
Until next time happy listening,