I am looking to install a new SSD on my system to run the OS, DAW and store some sound samples.

Researching the latest makes & models, one point of difference seems to be between drives using synchronous, and drives using asynchronous memory. This seems to boil down to their differing abilities to deal with 'compressible' and 'incompressible' data.

So my question is: from an SSD's point of view, are the files processed by a modern DAW considered to be compressible or incompressible data?

WAV files can be compressed without losing info so I presume the answer is that they are compressible.

no. You could compress it (try!) but the resulting file would be about the same size. The reason for this, is that compressing works if there is a lot of repeated data in your file. This is not the case in a normal wav file (you could see it as a big collection of numbers that are not so well related). MP3 compression is much more clever (so much so that I have no clue how it works ) but changes the audio.

--------------------
Music Chromelegs
Gear Rhodes | Cp70 | Nord Electro

I should have clarified when talking about WAV file as I was referring to WAV files containing uncompressed PCM data as that is what I assume we are talking about here. WAV files can contain other data types including compressed formats.

Quote Stage73:

---

no. You could compress it (try!) but the resulting file would be about the same size. The reason for this, is that compressing works if there is a lot of repeated data in your file. This is not the case in a normal wav file (you could see it as a big collection of numbers that are not so well related).

---

I just compressed a 42MB WAV file of a small piece of classical music using 7-zip and the resulting file is 26MB.

Quote Agharta:

---

I just compressed a 42MB WAV file of a small piece of classical music using 7-zip and the resulting file is 26MB.

---

Yes, I confirm this is possible with PCM-encoded audio. Using WinRAR or similar softwares, if you choose "best" in the "compression method" list then you can achieve a compression ratio of about 50% (and, of course way better ratios if the file contains lots of digital silence).

http://en.wikipedia.org/wiki/Run-length_encoding

The lossless formats like FLAC, APE and MLP use should use a similar technique (with some enhancements most probably, like clever multitrack management, etc).

Quote Stage73:

---

MP3 compression is much more clever (so much so that I have no clue how it works ) but changes the audio.

---

As far as I remember, this is more or less like this:

Encoding
0. Optional removing of very low or very high frequencies
1. Spectral analysis
2. Suppression of all bands that are not loud enough
3. Spectral data resolution reduction (actually, the bands that are "psycho-acoustically less relevant" have an even lower resolution)

Decoding
1. Resynthesis from the data obtained at the end of the encoding


So with MPEG-3 you lose:

- very low level audio
- extreme frequencies
- spectrum resolution
- transient clarity (because analysis/resynthesis affects this quite a lot)

thanks Bob for the mp3 explanation. I also tried zipping a wave file on my mac, all default settings, file goes from 31MB to 30.5MB. I have no clue how 'compressibility" would affect a hard drive performance (sounds like the drive is doing "secret things" with the data, haha)

--------------------
Music Chromelegs
Gear Rhodes | Cp70 | Nord Electro

Just as a point of interest does anyone know if a fft would be smaller or larger than the original pcm data?

--------------------
Daniel Davis
Edinburgh Recording Studio Windmill Sound

That depends on the windowing size of the FFT?

--------------------
~~~ A weasel hath not such a deal of spleen as you are tossed with! www.feline1.co.uk ~~~

I don't really understand the question of the FFT. FFT is an algorithm, not a computer file description. If you have a music recording in WAV format, it is just a sequence of numbers that essentially tells the loudspeaker (via your interface and amplifier) cone how to move (max value: push speaker outwards, min value pull speaker inwards).

Now if you look at these numbers in a graph, you will see the waveform (describing what the speaker cone does). Most waveforms look more or less like a alternating series of mountains and valleys, you could see it as a "dirty version"of a sine wave. It happens to be so, that you can describe complex waveforms by adding up and subtracting series of (a lot of) perfect sine waves. This is called a fourier transform: basically it transforms your music signal into "graphic equalizer bands jumping up and down". FFT, fast-fourier-transform, is an invention that does this so efficiently, it can even be carried out in a chip or in your DAW in real time.

With this info, your computer could move the speaker by saying: play me a sine wave of 10V at 200Hz, subtract a sine wave of 300Hz of 1V, and add another sine wave of 0.1V at 600Hz, as if you would build your music signal up again from the infos of your jumping equalizer (in this case, we would have 3 bands, first one with all lights on, second one with only a couple,...).

Here it gets difficult: if you use an FFT to break down your music into equalizer infos, the analysis needs a bit of time. Imagine you try to get frequencies out of only one point in time, one sample: that is not possible. You need a larger series of points, for example, 1000 samples. This is the time frame. Just as your graphic eq jumps up and down, the frequency content changes with every time frame analyzed.

Instead of storing a waveform as a simple series of numbers, you could also define a time frame, and then store the jumping eq results from an FFT for each frame into a file. In a digital system, I think it is possible to recreate the waveform from that with 100% accuracy, but I am not sure (I am only a mechanical engineer hahaha). In any case, I don't think it is a more efficient way of storing the data, unless you start throwing away information on very high or very low frequencies. The fun of this FFT-ized audio, is that you can start manipulating the audio signal (equalizing)

wow....that is a lot of text....please if anybody knows more about digital signal processing jump in and correct me where I am wrong...as said, I am only a mechanical engineer so explanations have to be taken with care.

--------------------
Music Chromelegs
Gear Rhodes | Cp70 | Nord Electro

Quote Stage73:

---

thanks Bob for the mp3 explanation. I also tried zipping a wave file on my mac, all default settings, file goes from 31MB to 30.5MB. I have no clue how 'compressibility" would affect a hard drive performance (sounds like the drive is doing "secret things" with the data, haha)

---

On OSX, try SimplyRAR, it has the same options as WinRAR and can thus achieve very good compression ratios.

Using FLAC (free lossless compression), can compress to between 1/3 and 1/2 the original size depending on the material.

--------------------
It wasn't me!
(Well, actually, it probably was)

Why not?
There are so many choices for you
e.g. FLAC, Monkey's audio, Wavpack, TAK, TTA, Shorten, Apple lossless, WMA lossless

You should take a look to this:
http://en.wikipedia.org/wiki/Lossless_audio#Audio

--------------------
Cubase 5.1 + Foobar2000 -> ASIO + MMCSS + Windows7 -> SIIG TI chips 1394 PCIe -> Echo AudioFire 4 -> HH MX250 + Wharf Delta70
A native USA style contemporary R&B artist in HongKong

Quote Daniel Davis:

---

Just as a point of interest does anyone know if a fft would be smaller or larger than the original pcm data?

---

As feline1 suggested, it depends on the analysis parameters (window size, overlap, number of frequency bands, etc), and you can choose them so that it produces less data than the original PCM data. Then this data is further compressed by reducing its resolution, and by other clever stuff.

Btw I am not even sure that MP3 uses FFT. It does use spectral analysis and resynthesis, but there are many ways to do it.

Quote Stage73:

---

thanks Bob for the mp3 explanation.

---

That is indeed a good basic explanation from Bob. Actually I wrote a longer one about the tradeoffs between MP3 bitrate and audio quality and what to listen out for recently in SOS if anyone wants a little more detail

www.soundonsound.com/sos/jan12/articles/qand1-0112-2.htm

Hope it helps!


Martin

--------------------
YewTreeMagic

that is a nice article (I particularly like the way you recommend certain bitrates in practical terminology rather than scientific blabla for audiophiles)

however, we are straying a bit from the original question on go Tbird: how the hard disk handles compressible or incompressible data. This is more an IT-issue than a musical one (and has nothing to do with neither FFT nor MP3). I know hard disks have a clever way of storing data all around the disk and this affects the reading and writing speed, and food SSDs this might be different?

maybe somebody knows?

--------------------
Music Chromelegs
Gear Rhodes | Cp70 | Nord Electro

Interesting, I would have thought the term "compression" as it applies to data would mean a process that is totally reversable in that the reconsructed information, a document, photo or computer program, such as the compressed trials one can download, is exactly the same as the original file?

Are the various "lossless" formats of that type? I guess MP3 is a form of compression in that it makes things smaller but, (and I am sure it is far too late to do anything about it!)maybe it should be called "decimation"? But in the modern usage of the word, i.e. to remove a large proportion and not the orignal meaning of removing one part in ten.

Dave.

Dave, you are right to flag the often misleading use of the term 'data compression'. As you say, to audio types, 'compression' generally means a reduction in amplitude, but which can be undone by the equivalent 'expansion' -- as employed by noise reduction schemes such as Dolby's various offerings.

Data compression does exist -- systems like Winzip or PKzip provide the ability to 'compress' the data to a smaller size in such a way that it can later be 'expanded' to regenerate the original bit-perfect file. This is a 'loss-less' system.

In audio terms, there are digital audio file formats that provide loss-less data compression: FLAC, Apple loss-less, MLP, and many others.

But there are many more that are 'lossy' -- such as MP3, ATRAC, AC3 and so on. These are 'data reduction' schemes because they reduce the file size, but in a destructive way which can not be undone.

The terms 'audio data compression' and 'audio data reduction' are often used interchangeably, but they do imply quite different things, and we endeavour to use them in their correct contexts in SOS.

hugh

--------------------
Technical Editor, Sound On Sound

Quote TBird:

---

I am looking to install a new SSD on my system to run the OS, DAW and store some sound samples. Researching the latest makes & models, one point of difference seems to be between drives using synchronous, and drives using asynchronous memory. This seems to boil down to their differing abilities to deal with 'compressible' and 'incompressible' data.
So my question is: from an SSD's point of view, are the files processed by a modern DAW considered to be compressible or incompressible data?

---

This thread has gone off topic somewhat.
If your DAW of choice saves its audio data as ‘regular’ PCM WAV files then that data is considered as being compressible by the SandForce controller that your query relates to.
If your DAW uses a proprietary file format to save audio data to it is possible that it could use a lossless data compression technique so you’d best check with the manufacturer.
My guess is that DAWs will not typically use lossless data compression for user recorded audio data but you can test if the files are compressible by using WinZip or 7-zip etc. If you can compress those files using such utilities then the SandForce controller can also compress them and you will get better performance.

I know I am a bit of a computer numpty but I had gained the impression that SSDrives were at their most useful as system drives, giving faster boot times?

Music files were best stored on a second physical drive and a modern SATA drive was fast enough for this purpose? If I am right then I cannot see any merit in compressing .wav files at all (unless you are 24bits at 192kHz nutter!). I know we have come through a price blip but conventional drive prices are again quite low. Indeed a gander at Maplin shows a 1.5TB usb 3.0 drive at 95 quid!

Dave.

MP3 and its related data reduction schemes use the discrete cosine transform rather then the fast fourier transform but it amounts to very much the same thing, some of the more modern schemes use wavelet transforms instead which are conceptually more complex but can manage better compression before the artefacts become objectionable.
The chapter on the Wavelet transform in "Numerical recipes" is quite a good introduction.

PCM audio generally compresses poorly using run of the mill Huffman or LZ schemes, but you can get 2:1 or so using audio specific data compression routines, I doubt that the sandforce part will implement these.
Besides a disk controller that advertises numbers assuming compressible data is cheating in my view.

No DAW will be compressing its internal data, it is too computationally expensive, but storing sample libs lossless compressed is reasonable (Decompression is typically **MUCH** cheaper then compression, and some of those datasets are huge).

Regards, Dan.

--------------------
Audiophiles use phono leads because they are unbalanced people!

Thanks agharta for bringing the thread back on topic!

--------------------
Music Chromelegs
Gear Rhodes | Cp70 | Nord Electro

Thank you everyone for your input re WAV files but I think I may have asked the wrong question; or rather, two subtly different questions in the same thread.

Kudos to Agharta for paying attention. Your answer is the most relevant.

Perhaps what I really meant to ask was -

"Does anyone have any practical experience of using solid state drives with their DAW and can they recommend any particular make & model?"

Or is that a new thread?

Quote TBird:

---

Does anyone have any practical experience of using solid state drives with their DAW and can they recommend any particular make & model?

---

I imagine this topic has come up before so have you tried searching the forums?

There is an article that benchmarks reading WAV files, converting them to MP3 and writing them to disk that looks at a number of SSDs:

http://www.myce.com/review/ocz-octane-ssd-review-57979/MyCE-Reality-Suite- 9/

The SandForce drives do well but are just beaten by the Indilinx controller. It’s hard to extrapolate from that as to how a DAW will benefit.

> No DAW will be compressing its internal data, it is too computationally expensive

Cubase and others do this (Cubase lets the user use FLAC).

> "Does anyone have any practical experience of using solid state drives with their DAW and can they recommend any particular make & model?"

Crucial C300 (now sadly discontinued) although it depends on WHY you want to use an SSD.

I have the C300 (128GB) and it works very well for what I do. I chose it because it was the fastest I could find for reading small files. Some SSDs are better for big files, etc... reading SSD benchmarks is not that easy, and I just forgot everything about it.

That said, I only use the SSD for the operating system and the programs (which is why I was interested in reading small files very fast). All data is stored on a regular hard drive (Western Digital Black 1TB), including all audio files. Big SSDs are still expensive! But if you can afford one, they are even faster than small SSDs -- I have no idea why.