Help for a newbee

[ben595]

Hi everyone,

I'm a newbee in the world of Hi-Fi audio and I'm a bit confused.

1. When I read posts on this site, I can see words like "pre-amp", "DAC" and "NOS DAC".

I search the net and foud out that DAC is for "digital to analog converter" but I'm not sure what it is used for.

If I plan to make a gainclone, when do I need one of these (supposing they are not all reffering to the same thing)?

It I understand correctly, a DAC will be needed if I plan to connect a commercial CD player to my gainclone. If I plan to connect my gainclone to my computer or a Ipod, I don't need anyting else since those two devices have already converted the audio signal. Correct?

2. If I want to listen to MP3 from my computer, how can something enhance the sound between the computer and the gainclone? The computer already "decode" the file. Is there something left to "decode"?

Thank you very much

Ben

 

[planet10]

A CD player has a DAC in it, so does your computer. An external DAC is used because it is (most often) better than the built-in one.

dave

 

[nac134]

This is how the music gets from the mp3 file or cd to your ears

1. First of all, your computer (or other device) spits out a digital signal that consists of ones and zeros, which is called a "bitstream". This can travel over usb, a toslink optical digital cable, or a coaxial digital cable. The most important thing here is the bit rate - how many ones and zeros per second. More is better. 128 kbps is 128 kilobits per second - so more kbps equals better sound.

2. You then need a device to convert this digital signal to analog. This is the DAC. Computers, cd players, and other devices have these built in. Whenever you are using the stereo output from a device, you are using its built in DAC. When you use a digital output from a device, you are not.
The quality of the DAC has a large effect on sound quality. Computers, ipods, and low cost electronics usually have low quality DAC's, so you can improve the sound by using a higher quality external DAC. An external DAC will have a digital input, which will come through a single cable, and an analog output, which will use 2 cables for stereo.
Some DACs boost the bitrate to try to improve the sound. This is called oversampling. NOS is non-oversampling, which is popular lately. Which is better is a matter of personal preference.

3. Once you have an analog signal coming out of your DAC, you need something to control the volume level. This is the pre amp. Pre amps typically also allow you to switch between different sources, ie cd player, tape player, etc., but some only work with one.

4. The signal coming out of your preamp goes into your power amp, in this case your gainclone, which provides power to your speakers. You can also make the pre amp and power amp into one piece - this is called an integrated amp. It is very easy to build a gainclone with a built in pre amp.

5. Anything that has its own volume control can be run into a power amp directly. Anything without volume contol needs some kind of pre amp.

A good way to proceed would be to build an integrated gainclone (a gainclone with volume control). It will work with your pc, ipod, or cd player. After that, you can buy or build a high quality DAC which will significantly improve the sound.

 

[ben595]

Wow! Thanks for the complete reply.

So, if I understand correctly, if I want to digitalize my CDs and have the best sound possible given this source, I should buy a sound card with a digital out, build a DAC and connect it to my Gainclone with a volume control integrated. The soud quality I'll get will be more dependable to the mp3 quality than that of the hardware.

Since the Digital out is a bitstream, can I buy a cheap sound card with a digital out or is it better to buy a good one like a M Audio for 100$?

As for the bitrate, how many kbps is rated as a great quality? 128, 192, 320kbps or even higher? Will I see any difference?

Thanks again

 

[planet10]

if I want to digitalize my CDs and have the best sound possible given this source, I should buy a sound card with a digital out, build a DAC and connect it to my Gainclone with a volume control integrated. The soud quality I'll get will be more dependable to the mp3 quality than that of the hardware.

If you want the best sound DO NOT import the CDs into MP3s, but full resolution AIFF, WAV, FLAC, or ALAC files.

And you can also consider no sound card at all, and get a Firewire (my preference) or an Asychronous USB DAC (or USB to SPDIF convertor)

dave

 

[Slipstreem]

As for the bitrate, how many kbps is rated as a great quality? 128, 192, 320kbps or even higher?

If you're going to convert your CDs to MP3 then it's worth making a lossless backup first as suggested above. This way, you'll never need to rip your CDs again and can store them away safely as backups.

There is no reason for there to be any audible difference between a lossy MP3 encoding made properly and the original lossless file, but you can't reconstruct the original file from an MP3 file, hence the importance of a first-stage lossless rip.

People who understand how MP3 works seldom ever talk in terms of bitrate nowadays as VBR (variable bitrate) is capable of producing better results for a given file size than any CBR (constant bitrate) encoding, with the exception of CBR320, so VBR is almost always used in preference to CBR encoding.

VBR has been the main focus of most MP3 encoder fine-tuning development work for many years now, with the LAME encoder used at a setting of -V2 being considered widely as the de-facto standard. This gives files that average roughly 190kbps but allows the encoder to intelligently select any bitrate between 32kbps and 320kbps on-the-fly many times per second during the encoding process.

Transparency is never guaranteed with lossy encoding, but you would have to have very keen ears indeed to be able to tell a LAME -V2 encoding from the original lossless file in a blind comparison test. The vast majority of those who do claim to hear a difference are victims of placebo effect and can't tell them apart in reality. The tiny minority of those who genuinely can tell the difference are almost always satisfied by a -V0 encoding instead, although these will be significantly larger, averaging around 245kbps.

Remember, you can never accurately predict the size of a VBR encoding as the bitrate varies dynamically to suit the complexity of the content fed to the encoder, although being able to predict the exact final file size is seldom relevant unless you're seriously short of storage space. CBR = constantly varying quality with a predictable file size. VBR = constant quality with a slightly unpredictable file size.

For further information on using LAME to make MP3 encodings, please refer HERE. This particular article is maintained by those involved in LAME development. If you find any other sources of information regarding LAME that contradict the information given in the official WIKI (and there is plenty out there), it's just plain wrong, so ignore it.

Will I see any difference?

We listen with our ears, not our eyes, so no.

 

[planet10]

There is no reason for there to be any audible difference between a lossy MP3 encoding made properly and the original lossless file

I beg to differ.

dave

 

[Slipstreem]

OK. Provide some evidence then. A log file from a personal ABX test comparing lossless to a LAME -V0 encoding would suffice.

 

[nac134]

Even full CD quality is not as good as it gets. SACDs have even higher resolution, and sound better than CDs. I agree with planet10 - you should import your CDs in a lossless format. I also agree with his (or her) suggestion that you use a USB to SPDIF converter. This allows you to easily connect an external DAC to your computer, and they are not too expensive. Keep in mind that sound quality differences from different formats may not be readily apparent with modest sound systems. As you move to higher end systems, these small differences become more important. MP3s cannot match lossless encoding. On my stereo one could easily tell the difference ten times out of ten.

 

[Slipstreem]

MP3s cannot match lossless encoding. On my stereo one could easily tell the difference ten times out of ten.

OK. As you're happy to quote "ten times out of ten", this infers that you must have scientifically tested this claim either with DBT or ABX tests. Let's see the results then please.

 

[nac134]

Neither ABX or DBT are scientifically valid methods.

 

[Slipstreem]

Says who?

 

[nac134]

Anybody who does actual science. Imagine doing proper chi-square testing or ANOVA on either of these tests. The implications of qualia theory vs. representation theory. The high likelihood of myriad lurking variables. To get the necessary p < 0.005, you would need a prohibitively large sample size, and even then the results could be strongly challenged from multiple angles.

 

[Slipstreem]

So, converting the above to layman's terms for the benefit of most people reading this, are you saying that not being able to hear a difference doesn't mean that you can't hear a difference, assuming that the same hardware is used for testing as is used for general listening?

 

[nac134]

You asked me if my claims were scientifically tested. If you don't understand the above terms, which cannot really be reduced to layman's terms, you don't know what scientific testing is. Thats not an insult - most people actually know very little about how science really works. You pretty much need to study science in university to understand what is or is not scientifically valid. And normal AB testing in not scientifically valid, though I would not call it useless.

 

[Slipstreem]

I do understand the above terms (loosely), but I'm sure there will be many reading this who don't. That's all I was saying.

All I was trying to get from you was a simple 'yes' or 'no' answer as to whether you considered a controlled test that can prove to an individual whether or not they can hear a difference between two sources had any validity.

Whether or not you class it as scientifically valid, it doesn't alter the fact that a properly conducted ABX test will tell an individual whether or not they can actually hear a difference between two sources, or whether placebo effect has clouded their judgement.

The advice that's generally given to 'knock it up a notch' above the borderline setting that reaches perceptual transparency for a given individual in terms of choosing an encoding bitrate/VBR quality setting to use will generally cover a person when it comes to harder-to-encode material. It's not a catch-all, but it's definitely a catch-most.

Beyond this point, I fail to see why a layman should have to worry himself about statistical likelihoods if almost everything he encodes is almost guaranteed to be perceptually transparent to him. After all, if a person wants 100% guaranteed perceptual transparency at all times, they won't be lossy encoding in the first place.

 

[planet10]

Neither ABX or DBT are scientifically valid methods.

ABX is a specific test, and very often misused or misinterpreted. It can never prove that anything is indistinguishable. One person repeatably being able to hear a difference is sufficient to show that there is a difference.

DBT on the other hand is a whole class of experiemnts and some are well done & valid, some aren't.

dave

 

[planet10]

All I was trying to get from you was a simple 'yes' or 'no' answer as to whether you considered a controlled test that can prove to an individual whether or not they can hear a difference between two sources had any validity.

It is possible for such a test to show that someone can hear a difference. If no differences are heard results are only valid for that session (equipment, room, stimulaus & state-of-mind)

One of the big issues with these tests is that when you concentrate to pick out differences, you use different parts of the brain to analyse what you hear.

Also in an ABX test, ethe number of trials required to get a statistically valid result, introduce a strong possibility of subject fatigue, which in itself can invalidate the results.

dave

 

[planet10]

One person repeatably being able to hear a difference is sufficient to show that there is a difference.

A good example of that is a compression standard that a standards body was working on (Eropean IIRC). Many many DBT trials with 100s of particiapants.

Just before ratification, BN Locanthi has a listen and points out a clearly audiable "whistle". They had to start over.

dave