[wyager]

Quote:

Quote:

Think about it like this. Let's say that the maximum voltage your ipod can produce is 1.27 volts (as we learned might be more accurate). This means that, with an 8 bit audio file, the minimum voltage change is .01v. Now, if we used a 16 bit audio file instead, the sample rate stays the same (44100hz) and the max voltage stays the same, but the minimum voltage change is now .0000390625v, which is a much more accurate reproduction of the real sound than 8 bit audio can provide.

 
Well, that's the question, if it really works that way ... and shouldn't we consider the kind of DAC used, 16 bit or 8 bit or even 24 bit?
 
Your simple mathematics seem right (well, see below), it's also almost the same as mine in my first posts here.
But I'm not sure anymore if it is really handled that way by the ADC/DAC and the other components analog and digital involved.
Also I have the impression as if a 24 bit file on my 801 sounds more loud than a 16 bit file with the same volume setting.
 
And what happens if you plug an amplifer into your ipod which is capable of 5 volts or more (... and let*s not forget playing such a song via a dektop amplifier and 200 watt speakers ...)... would it just thin the sound 

?
 
(... 8 bit = 256 values ... 1.27 / 256 = 0.005 ... or do I miss something ?)

That's true, the kind of DAC used may affect the output. You say that a 24 bit file sounds louder-but does it sound 256 times louder? That would mean it is acting as the OP says. And yes, if that amp increased the volume then I see no reason why the steps wouldn't be even more noticeable. And yes, for your last equation you are forgetting that the (hypothetical, but realistic) voltage range is actually +1.27 to -1.27, which makes for a total difference of 2.54 volts or .01v per step. 
 
I'm not sure if this will make sense to anyone here, but the way I think of it whichever device you're using just puts the MSB of your 24/16/8/whatever bits in the MSB of the DAC's output register and makes the rest zeros. The way the original post says it would work is if the LSB of the audio file was matched to the LSB of the DAC (which seems unrealistic). So, let's say that your DAC is some 8 bit microcontroller (as that is both likely and simple to match up to 24/16/8 bit audio), and the song you are listening to is a 24 bit audio file. The current voltage in 24 bit binary could be 10101010, 10101010, 10101010 (.666666627v). With the the way I believe it works, if this file was downgraded to an 8 bit file, the first 8 bits of the original audio file would be copied and the binary value in the DAC would be 10101010, 00000000, 00000000 (.6640652v). As you can see, the volume change would be minimal (with lower quality audio). The way the OP is suggesting, the value instead would be 00000000, 00000000, 10101010, or about .00001013v. The volume in this case would be 65,536 times lower than the original 24 bit audio file, with no advantages in terms of processing power or even quality. It just doesn't make sense to me that anyone would choose option 2. Also, sorry if that was confusing, that just makes sense to me from a low-level software standpoint.

 

[Lenni]

I've not read the most of the thread as it gets too technical, but if it's of interest I've uploaded two flac files of the same song, for educational purpose (I hope is not against the rules).
 
one is a Studio Master 24bit 96kHz from Linn Records
 
the other is a CD rip
 
the difference between these two file is quite drastic. one is indeed much louder than the other. but hear it for yourself link
 
note: if for some reason the sharing of these file is not permitted, even for educational purpose, and you're interested in hearing the files, pm and I'll pm back the link.

 

[ramicio]

You won't convince people that way because mastering is quite different between the CD and HD tracks.  I still believe and have heard differences between 2 equally mastered 16/44.1 and 24/96 files by simply taking a 24/96 file and down
sampling and dithering down the bit depth.  It's not drastic, but as I said before, we have the storage technology, so why not increase quality?  Everything else in the world has gone to increasing quality.  Hell most people who think 16/44.1 is enough also think you can't tell the difference between a 128 kbps AAC and the uncompressed source.  We used to have lossy audio on DVD, then another storage format came along and just multiplied the image resolution by 6 and we gained better video compression, so now we have room for lossless audio. They did it, why can't the music industry.  It would help protect them from theft because the majority of dumb people out there don't like dealing with large files over the internet and waiting.

 

[Lenni]

yes, the whole mp3 development was due low Internet speed and storage... this in not longer justifiable. however I struggle finding differences between mp3 and lossless these days myself.
 
btw I don't wish to convince anybody, I have the files and thought might be relevant to the thread  

 

[ramicio]

Lossy formats have stuck because the majority of people are going to internet downloads, either through purchase through places like iTunes or illegally.  Lossy is in the interest of the providers because it saves bandwidth.  The source end of anything could give 2 craps about the quality of the production and format as long as it sells.  The only way music seems to sound good anymore is if the artist cares at all, has their own label, or is an unpopular type of music.  Just about every artist with their own label are rapper entrepreneurs, so all they care about is money anyway, and it's part of pop so it's going to be loud.  I just think MP3 was one of the worst codecs as far as bitrate vs. quality.  Everyone has always said 128 kbps MP3 was transparent to CD audio, then when AAC came out it was 128 kbps with that.  I don't care for lossy at all and I can hear differences between lossy and lossless.  I only ever use lossy for portability and non-quiet environments.

 

[xabu]

Quote:

Quote:

Quote:

Think about it like this. Let's say that the maximum voltage your ipod can produce is 1.27 volts (as we learned might be more accurate). This means that, with an 8 bit audio file, the minimum voltage change is .01v. Now, if we used a 16 bit audio file instead, the sample rate stays the same (44100hz) and the max voltage stays the same, but the minimum voltage change is now .0000390625v, which is a much more accurate reproduction of the real sound than 8 bit audio can provide.

 
Well, that's the question, if it really works that way ... and shouldn't we consider the kind of DAC used, 16 bit or 8 bit or even 24 bit?
 
Your simple mathematics seem right (well, see below), it's also almost the same as mine in my first posts here.
But I'm not sure anymore if it is really handled that way by the ADC/DAC and the other components analog and digital involved.
Also I have the impression as if a 24 bit file on my 801 sounds more loud than a 16 bit file with the same volume setting.
 
And what happens if you plug an amplifer into your ipod which is capable of 5 volts or more (... and let*s not forget playing such a song via a dektop amplifier and 200 watt speakers ...)... would it just thin the sound 

?
 
(... 8 bit = 256 values ... 1.27 / 256 = 0.005 ... or do I miss something ?)

That's true, the kind of DAC used may affect the output. You say that a 24 bit file sounds louder-but does it sound 256 times louder?
 
No and it doesn't need to, if one accepted that the gain is only 6 dB per bit, so this would be 48 dB more. E.g. at 1 kHz you have a doubling of sensed volume per 10 dB increase (which translates to 10 phon for every frequency)
But it also doesn't had to be 4 - 5 times louder because other aspects like differnt mastering might apply ... and perhaps there is still something wrong with the whole notion regarding the amped volume.
 
That would mean it is acting as the OP says. And yes, if that amp increased the volume then I see no reason why the steps wouldn't be even more noticeable. And yes, for your last equation you are forgetting that the (hypothetical, but realistic) voltage range is actually +1.27 to -1.27, which makes for a total difference of 2.54 volts or .01v per step. 
 
O.k., thought you meant the overall range is 1.27 v.
But if you were right und the sound would also change in it's structure apart from just getting louder then this had to be the case with all sound, not only your stepped sine.
This would really be one reason more to prefer headphones to speakers

 
I'm not sure if this will make sense to anyone here, but the way I think of it whichever device you're using just puts the MSB of your 24/16/8/whatever bits in the MSB of the DAC's output register and makes the rest zeros. The way the original post says it would work is if the LSB of the audio file was matched to the LSB of the DAC (which seems unrealistic). So, let's say that your DAC is some 8 bit microcontroller (as that is both likely and simple to match up to 24/16/8 bit audio), and the song you are listening to is a 24 bit audio file. The current voltage in 24 bit binary could be 10101010, 10101010, 10101010 (.666666627v). With the the way I believe it works, if this file was downgraded to an 8 bit file, the first 8 bits of the original audio file would be copied and the binary value in the DAC would be 10101010, 00000000, 00000000 (.6640652v). As you can see, the volume change would be minimal (with lower quality audio). The way the OP is suggesting, the value instead would be 00000000, 00000000, 10101010, or about .00001013v. The volume in this case would be 65,536 times lower than the original 24 bit audio file, with no advantages in terms of processing power or even quality. It just doesn't make sense to me that anyone would choose option 2. Also, sorry if that was confusing, that just makes sense to me from a low-level software standpoint.
 
We can believe this and that

but that wouldn't necessarily be the way it really works. Right at the moment I still have not enough information to get the complete context right.

 
My comments in bold above.
 
 
Removed my thoughts ... not completely coherent yet

, but I get it better and better ...

 

[wyager]

Quote:

 
No and it doesn't need to, if one accepted that the gain is only 6 dB per bit, so this would be 48 dB more. E.g. at 1 kHz you have a doubling of sensed volume per 10 dB increase (which translates to 10 phon for every frequency)
But it also doesn't had to be 4 - 5 times louder because other aspects like differnt mastering might apply ... and perhaps there is still something wrong with the whole notion regarding the amped volume.
 
OK, so does it sound anywhere near 48dB louder? Since decibels are a logarithmic unit, I believe that would be roughly 63,000 times louder. (even after the logarithmic compensation of your ear in the inverse direction, it would still be many times louder). That still seems ridiculous to me...
 
O.k., thought you meant the overall range is 1.27 v.
But if you were right und the sound would also change in it's structure apart from just getting louder then this had to be the case with all sound, not only your stepped sine.
This would really be one reason more to prefer headphones to speakers

 
I'm not quite sure I get this... Yes, increasing the bit depth would make all sounds sharper and clearer.
We can believe this and that

but that wouldn't necessarily be the way it really works. Right at the moment I still have not enough information to get the complete context right.
 
Sure, that's true. I'm just operating on what I've learned about DACs and sound output from a low level/hardware perspective. I can see no logical reason why anyone would make a device behave the way described here.

 

 

[xabu]

Quote:

Quote:

 
No and it doesn't need to, if one accepted that the gain is only 6 dB per bit, so this would be 48 dB more. E.g. at 1 kHz you have a doubling of sensed volume per 10 dB increase (which translates to 10 phon for every frequency)
But it also doesn't had to be 4 - 5 times louder because other aspects like differnt mastering might apply ... and perhaps there is still something wrong with the whole notion regarding the amped volume.
 
OK, so does it sound anywhere near 48dB louder? Since decibels are a logarithmic unit, I believe that would be roughly 63,000 times louder. (even after the logarithmic compensation of your ear in the inverse direction, it would still be many times louder). That still seems ridiculous to me...
 
As I said you have a doubling of sensed volume per 10 phon (10 dB at 1 kHz) ... read it up; so it's not 63000 times louder. The frequency, dB, phon, SP relations are a bit more complicated. The phon unit is already corrected over the frequency range based on our hearing ability.
 
O.k., thought you meant the overall range is 1.27 v.
But if you were right und the sound would also change in it's structure apart from just getting louder then this had to be the case with all sound, not only your stepped sine.
This would really be one reason more to prefer headphones to speakers

 
I'm not quite sure I get this... Yes, increasing the bit depth would make all sounds sharper and clearer.
 
This was about amplifying, not about more bits. Because you wrote that you would notice the steps even more with more amplifying (and I assumed you meant not only the volume but the degrading sound effect.)
 
We can believe this and that

but that wouldn't necessarily be the way it really works. Right at the moment I still have not enough information to get the complete context right.
 
Sure, that's true. I'm just operating on what I've learned about DACs and sound output from a low level/hardware perspective. I can see no logical reason why anyone would make a device behave the way described here.
 
Because it works obviously and millions of people are listening to such devices all day enjoying their music

. Read about dithering etc. and have a look at this sampling theory pdf I linked.

 

 

[D. Lundberg]

 
Quote:

Quote:

No and it doesn't need to, if one accepted that the gain is only 6 dB per bit, so this would be 48 dB more. E.g. at 1 kHz you have a doubling of sensed volume per 10 dB increase (which translates to 10 phon for every frequency)
But it also doesn't had to be 4 - 5 times louder because other aspects like differnt mastering might apply ... and perhaps there is still something wrong with the whole notion regarding the amped volume.
 
OK, so does it sound anywhere near 48dB louder? Since decibels are a logarithmic unit, I believe that would be roughly 63,000 times louder. (even after the logarithmic compensation of your ear in the inverse direction, it would still be many times louder). That still seems ridiculous to me...
 

0dBFS is the highest possible peak for a digital signal, regardless of bit depth. And at the other end you have the noise floor, which works as a (gradual) limit.
So a 16 bit signal can be just as loud as a 24 bit signal, the difference is in the dynamic range.
 
Quote:

I'm not quite sure I get this... Yes, increasing the bit depth would make all sounds sharper and clearer.

Higher bit depth will just lower the noise floor, so it won't make a difference unless the noise floor is high enough to be audible and/or mask parts of the signal.
There are no "steps" in a properly reconstructed signal, just a band limited (and relatively accurate) version of the original signal and some background noise.

 

[wyager]

 

[size=medium]

Quote:

 
Quote:

Quote:

No and it doesn't need to, if one accepted that the gain is only 6 dB per bit, so this would be 48 dB more. E.g. at 1 kHz you have a doubling of sensed volume per 10 dB increase (which translates to 10 phon for every frequency)
But it also doesn't had to be 4 - 5 times louder because other aspects like differnt mastering might apply ... and perhaps there is still something wrong with the whole notion regarding the amped volume.
 
OK, so does it sound anywhere near 48dB louder? Since decibels are a logarithmic unit, I believe that would be roughly 63,000 times louder. (even after the logarithmic compensation of your ear in the inverse direction, it would still be many times louder). That still seems ridiculous to me...
 

0dBFS is the highest possible peak for a digital signal, regardless of bit depth. And at the other end you have the noise floor, which works as a (gradual) limit.
So a 16 bit signal can be just as loud as a 24 bit signal, the difference is in the dynamic range.
I can agree with this... the problem was, the OP defined "dynamic range" as a difference, not a ratio... Now, thanks to wikipedia, I know that it's really a ratio! So yes, I agree with you. 
Quote:

I'm not quite sure I get this... Yes, increasing the bit depth would make all sounds sharper and clearer.

Higher bit depth will just lower the noise floor, so it won't make a difference unless the noise floor is high enough to be audible and/or mask parts of the signal.
There are no "steps" in a properly reconstructed signal, just a band limited (and relatively accurate) version of the original signal and some background noise.
Increasing the dynamic range also decreases the difference between steps... and as I said, I have, in experimentation, listened to an 8 bit audio signal and even "properly reconstructed" the voltage jumps were noticeable.

 ​

[/size]

Quote:

Quote:

Quote:

 
As I said you have a doubling of sensed volume per 10 phon (10 dB at 1 kHz) ... read it up; so it's not 63000 times louder. The frequency, dB, phon, SP relations are a bit more complicated. The phon unit is already corrected over the frequency range based on our hearing ability.
 
OK, cool. I haven't learned enough about sound units and such yet. Still, even after converting to Db/Phon, the difference would be very marked.
 
This was about amplifying, not about more bits. Because you wrote that you would notice the steps even more with more amplifying (and I assumed you meant not only the volume but the degrading sound effect.)
 
If you turn the volume up (ie use an amp) then the steps will be easier to detect. The same is true for all sound, you notice more detail at a higher volume.
 
Because it works obviously and millions of people are listening to such devices all day enjoying their music

. Read about dithering etc. and have a look at this sampling theory pdf I linked.
 
What works? Again, I see no reason from a design perspective to make a DAC work this way with 8 vs 16 vs 24 bit audio. And I know all about dithering and sampling theory, they really don't have all that much of an affect on my argument. 

 

 

[xabu]

Well, my main interest is in unerstanding how it actually works in today applications and systems and not how it could be realized differently.
The thread was not about how 24 bits could be used in another way as they are used in current systems, but that with the way they are used there is not that kind of difference in sound proclaimed by many.
And if your idea of how it should be done would lead to an overall better end result is not yet proven.

 

[wyager]

It's really proven IMO... dynamic range stays the same, overall range stays the same... Unless you think that lack of volume is a good thing, I fail to see how it could possibly be any better to adopt a system where less bits=less volume. 

 

[xabu]

Quote:

It's really proven IMO... dynamic range stays the same, overall range stays the same... Unless you think that lack of volume is a good thing, I fail to see how it could possibly be any better to adopt a system where less bits=less volume. 

I meant proven by applying it in a realworld system, not theorethical. Build such a system and if it sounds much better than existing ones you will find your market.

 

[wyager]

What I'm saying is that I think "real world" systems DO work the way I've described... so far, none of my testing has suggested that the way you seem to be saying these things work IS the way they work... In fact, looking at sites where people actually make sound and don't just listen to it (like stackoverflow) I see nothing to suggest that there is a volume difference between 16 and 8 bit. So again, what I'm saying isn't theoretical, I believe it's factual and I haven't heard any convincing evidence otherwise.

 

[xabu]

Well, seems we have a problem here Houston

 
 
It's not aboute volume it's about dynamic range. Whatever, we go in circles ... or nowhere (by the way it's interesting that the word "nowhere" can be fragmented as "now here" which gives it a complete new sense, by itself and in relation to the original word)
 
So apart from our understanding or thinking or proving, it works somehow