Google Search  |  Meridian Systems  |  Meridian Streaming  |  Restaurant  |  WiKi  |  Duncan's Meridian Info  |  Board Rules  |  Restaurant Rules
Page 15 of 17 < 1 2 ... 13 14 15 16 17 >
Topic Options
Rate This Topic
#125098 - 2010-09-14 14:47 Re: Transporter Round Two (was Transporter for 999...) [Re: VirusKiller]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
Update:

I've been doing some more research on dither - the Wikipedia article is actually very good - I'm now completely convinced that dither need not be applied when simply adding 8-bits of zeros to 16-bit audio, because there is no re-quantization taking place.

However, it's less clear to me whether or not dither is required in general when upsampling. When upsampling, but remaining at a 16-bit word size, the samples are being modified and, as such, re-quantized; therefore dither must be applied, and indeed, sox automatically adds dither with the following command-line:

sox 16_44_input.FLAC -b16 -C0 16_88_output.FLAC rate -vMab 90.7 88200

But what about upsampling whilst simultaneously increasing word length to 24 bits. Sox does *not* apply dither automatically with this command-line*:

sox 16_44_input.FLAC -b24 -C0 24_88_output.FLAC rate -vMab 90.7 88200

I've been forcing high pass dither in this scenario:

sox 16_44_input.FLAC -b24 -C0 24_88_output.FLAC rate -vMab 90.7 88200 dither -S

Now, I'm sure this does no harm with a 24-bit signal where you're never going to hear the bottom few LSBs, but is what I'm doing unnecessary? I'm beginning to think it might be, but I can't quite put the logical argument in place. Is it that the extra 8 bits provide the extra accuracy to describe the re-quantization?

On a slightly related subject, I'm trying to figure out whether or not resolution enhancement is relevant when upsampling and increasing word size and, if so, how to apply it to the 16-bit signal prior to upsampling. I'm not convinced that it is or it isn't required, but given that the 24-bit samples will utilize all 24 bits after upsampling, I'm leaning on the side of "not necessary". As a corollary, is resolution enhancement a "relic" of the pre-upsampling world?

If one did want to perform resolution enhancement, I believe that the Wadia document provides all the pointers that are needed, though when it says that Wadia adds a 9-bit signal to the 16-bit audio, whilst correct, it's a little misleading if you don't have your wits about you. Bear in mind that 16-bit audio is +/- 15-bits with the 16-bit effectively being the sign bit. What Wadia is doing is adding a 9-bit *signed* signal - i.e. +/- 8-bits - to the original 16-bit audio. By adding (or subtracting) *8* bits into the "empty" 8-bits in the new 24-bit word, it will never overflow into the LSB of the original 16-bit word.

The final thing about resolution enhancement that I need to understand is how the (TPDF) amplitude distribution and the frequency distribution of the resolution enhancement combine to create the resolution enhancing signal.

Any thoughts greatly welcomed!



* To thicken the plot, if you specify "dither" as a DSP effect explicitly, the -a option invokes an algorithm which attempts to work out if dither is required or not. In this case, dither *is* applied automatically for both these command-lines:

sox 16_44_input.FLAC -b16 -C0 16_88_output.FLAC rate -vMab 90.7 88200 dither -a
sox 16_44_input.FLAC -b24 -C0 24_88_output.FLAC rate -vMab 90.7 88200 dither -a

The sox man page does suggest that this algorithm isn't perfect though.
_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
#125118 - 2010-09-15 02:11 Re: Transporter Round Two (was Transporter for 999...) [Re: VirusKiller]
ncpl Online   sleepy

President of the Imperial Galactic Government

Registered: 2002-02-14
Posts: 7133
Loc: Surrey, UK
VK, remind us....what was the 90.7 for ?
_________________________
Rgds,
Nick

Join the Last.fm Sooloos group and see what Hitchhikers listen to.

#15SE

Top
#125124 - 2010-09-15 02:34 Re: Transporter Round Two (was Transporter for 999...) [Re: ncpl]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
Reduces the bandwidth of the filter to 90.7% (or 20kHz). Basically, you can make the filter less "brick-wall" by starting to reduce the roll-off earlier. The sox filter bandwidth is, probably, the frequency below which the attenuation starts.

The other thing I forgot to mention last night is that I'm trying to find out the shape of the sox upsampling filter's frequency response. Specifically, does the filter response reach zero before the Nyquist frequency.

_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
#125142 - 2010-09-15 04:16 Re: Transporter Round Two (was Transporter for 999...) [Re: VirusKiller]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
As an aside, the impulse response that Ayre achieves in its "Listen" MP filter is really amazing and is achieved by rolling off the filter even earlier than 20kHz. Interestingly, the slow roll-off in the frequency response extends well beyond the Nyquist frequency (it is only -6 dB at 22.05kHz).


http://www.ayre.com/pdf/Ayre_MP_White_Paper.pdf

I'm not saying that this is superior to what Meridian has produced. As posted in another thread, it gets to a point where individual preferences count.

I'd like to have a go at replicating this, but I'm not sure that sox will allow this. EDIT: I think that the -a parameter "allow aliasing" must be the equivalent of the slow roll-off. Excellent!


Edited by VirusKiller (2010-09-15 04:28)
_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
#125144 - 2010-09-15 04:40 Re: Transporter Round Two (was Transporter for 999...) [Re: VirusKiller]
WelwynNick Offline
Hitchhiker

Registered: 2010-02-08
Posts: 338
Loc: Welwyn, Herts, UK
Everyone here has been critical of the Ayre apodising solution for some reason. A bit of not-invented-here creeping methinks. I think Ayre have a lot of the right answers with digital audio, including some that M don't have.

Nick

Top
#125145 - 2010-09-15 04:47 Re: Transporter Round Two (was Transporter for 999...) [Re: WelwynNick]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
There's an interesting article here which suggests that Ayre is using a slow roll-off (allowing aliasing) whereas Meridian does not allow aliasing (and therefore has more post ringing). I'm not sure if this is true or not, but it's an useful pointer.

BTW, I'm not critical of Ayre at all. On the contrary, it's refreshing to see another clearly very competent company have a go at all this. As I said above, it gets to a point where individual preferences are important.

Edit: Charles Hansen of Ayre has, of course, posted on this forum and made a very useful contribution: http://www.meridianunplugged.com/ubbthreads/ubbthreads.php?ubb=showflat&Number=95617#Post95617 and subsequent posts in that thread.


Edited by VirusKiller (2010-09-16 00:28)
_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
#125148 - 2010-09-15 04:58 Re: Transporter Round Two (was Transporter for 999...) [Re: WelwynNick]
Ronnie Offline
Ravenous Bugblatter Beast of Traal

Registered: 2001-01-08
Posts: 3757
Loc: UK
Originally Posted By: WelwynNick
Everyone here has been critical of the Ayre apodising solution for some reason. A bit of not-invented-here creeping methinks. I think Ayre have a lot of the right answers with digital audio, including some that M don't have.

I don't believe that most here have been in any way critical of Ayre's approach at all. And I don't remember many, if any, posts criticising their approach

Now, I think it's fair to say that many more here will have HEARD and liked Meridian's Apodizing approach and understand it better than Ayre's, but that's only because many here already own such equipment

But I haven't seen that criticism in any great volume, if at all, on here...........

Top
#125150 - 2010-09-15 05:19 Re: Transporter Round Two (was Transporter for 999...) [Re: Ronnie]
WelwynNick Offline
Hitchhiker

Registered: 2010-02-08
Posts: 338
Loc: Welwyn, Herts, UK
I can't find them now, but I recall a number of recurring comments suggesting Meridian's apodising solution was superior to Ayre's because it didn't have any pre-ringing; mis-informed things like that. I like Meridian's approach in many ways, and my 861 is head and shoulders above any other processor that I've heard, but I don't buy into everything.

For anyone who's interested in what Charles Hansen thinks (like me) I'd recommend thsi thread, which is long, but its a goldmine:

http://www.avsforum.com/avs-vb/showthread.php?t=1181755&highlight=ayre

Bob Stuart and Charles Hansen are my audio heros, but the difference between them is that Hansen is ever-so-slightly more forthcoming about what he thinks and what he wants to do.

Nick

Top
#125162 - 2010-09-15 07:39 Re: Transporter Round Two (was Transporter for 999...) [Re: WelwynNick]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
Originally Posted By: WelwynNick
I can't find them now, but I recall a number of recurring comments suggesting Meridian's apodising solution was superior to Ayre's because it didn't have any pre-ringing; mis-informed things like that.
There was one recent erroneous post to that effect which compared Meridian's apodizing to a previous (non-apodizing) generation of Ayre filter.

Thanks for the link; I'll take a good look at it.
_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
#125268 - 2010-09-16 09:08 Re: Transporter Round Two (was Transporter for 999...) [Re: VirusKiller]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
I've taken a look at the sox source code and it appears that the bandwidth parameter "-b" may represent the frequency response at -3dB roll-off, not the highest frequency that is not attenuated. So:

"-b 95.0", -3dB at 20.95kHz, translates into 20.53kHz, below which there is no attenuation.
"-b 90.7", -3dB at 20.00kHz, translates into 19.22kHz, below which there is no attenuation.
"-b 85.0", -3dB at 18.74kHz, translates into 17.49kHz, below which there is no attenuation.

The source code is very complex and completely undocumented, and I could be wrong. However, the proof of the pudding is in the eating, so here are some results. Let's call the filter "sox #4".

I was a bit worried about using the -b 85.0 filter as the unattenuated bandwidth appears pretty low, but this is what I get in sox when I use the lowest (85.0%) bandwidth setting allowable with *no* aliasing ("rate -v -M -b 85.0 88200 dither -S"):



And, assuming I got it right, here is the corresponding frequency response generated from white noise (which wasn't 100% flat below 1kHz):



Here is Meridian's 808.2 impulse response again (which has more ringing than mine):



And here is Ayre's minimum phase (MP) "Measure" filter, which looks very similar to mine for both impulse and frequency response. smile Strictly, I think that sox #4 has one or two less cycles of ringing than Ayre's.



Note that both the Ayre MP "Measure" and sox #4 filters are pure minimum phase and 100% apodizing, as they have a null frequency response at 22.05kHz. I assume Meridian's is too.

Here also is sox #2 again which is without aliasing, but more bandwidth (and therefore a steeper cut-off and more ringing):



I might have overdone it on bandwidth reduction with sox #4, and perhaps a bandwidth parameter somewhere in-between 85.0% and 90.7% - say 87.5% - might be a better choice. Only our ears will tell, of course.


Edited by VirusKiller (2010-09-16 09:43)
_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
#125270 - 2010-09-16 09:10 Re: Transporter Round Two (was Transporter for 999...) [Re: VirusKiller]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
Originally Posted By: VirusKiller
EDIT: I think that the -a parameter "allow aliasing" must be the equivalent of the slow roll-off. Excellent!
Not so excellent, in fact...

Without aliasing, the sox filter frequency response is zero before Nyquist (22.05kHz); with aliasing allowed, the sox filter frequency response is zero at around 24kHz, above Nyquist (hence the aliasing), but way below the 40kHz or so in Ayres "Listen" minimum phase filter:



My next challenge is to see if the sox source code can easily be modified to extend the roll-off in a manner akin to what Ayre has done.
_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
#125272 - 2010-09-16 09:19 Re: Transporter Round Two (was Transporter for 999...) [Re: VirusKiller]
VirusKiller Offline

Don't Panic

Registered: 2004-04-15
Posts: 10728
Loc: nr. Milton Keynes, UK
A quick upsample later with -b 87.5 yields sox #5. Here are all my apodizing filters without aliasing. As expected, sox #5 has approximately the same amount of ringing as the 808.2 filter.

sox #2 (bandwidth = 90.7% = -3dB@20.00kHz = -0dB@19.22kHz):

"-b 85.0", -3dB at 18.74kHz, translates into 17.49kHz, below which there is no attenuation.



sox #5 (bandwidth = 87.5% = -3dB@19.29kHz = -0dB@18.25kHz):



sox #4 (bandwidth = 90.7% = -3dB@18.74kHz = -0dB@17.49kHz):



Edited by VirusKiller (2010-09-16 09:25)
_________________________
Forum Moderator
I am #25 in the HH1 photo of fame.

Top
Page 15 of 17 < 1 2 ... 13 14 15 16 17 >

Moderator:  Carl, Duncs, Ludwig, ncpl, VirusKiller 
Who's Online
6 registered (Akimo, Zanderbarfly, JobSeeker, Dappad, Gianni, MI'), 15 Guests and 5 Spiders online.
Key: Admin, Global Mod, Mod
Newest Members
Dkon, Big T, RussellP, jjcarr, Greg N
3969 Registered Users
Top Posters (30 Days)
Ian 74
Ratbert 74
ncpl 65
Carl 62
ChrisLayerUK 60
Top Posters
VirusKiller 10728
Fiddler 8202
ncpl 7133
Carl 5965
Ian 5962
Forum Stats
3969 Members
24 Forums
20215 Topics
209396 Posts

Max Online: 136 @ 2013-09-06 15:24

Meridian  |  Media Centre  |  Support  |  Firmware Release Notes  |  RSS Systems  |  RSS Streaming  |  RSS Restaurant