1 bit recording

[Boot_me]

Some interesting links on 1 bit recording.
Just in Case Reaper heads that way someday.

The spec mostly in Japanese
http://www.acoust.rise.waseda.ac.jp/.../020917wsd.pdf


A description of it.
http://www.sel.sony.com/SEL/consumer/dsd/dsd.pdf


Some audio comparisons...description of recording and convert down at bottom.
http://www.wingfieldaudio.com/portab...d-samples.html

A couple of guys discussing the quality of it and various equipment that currently supports it (back in 2003 and 2004).
http://www.gearslutz.com/board/remot...its-glory.html

A very good article on the Korg MR1...which is what I'm considering getting.
http://digitalmedia.oreilly.com/pub/...er.html?page=1

Anyway, here's some good information on it, pretty much all I could find after several hours data mining with Google.
Does anyone know of any normally priced Daw's that support multi tracking these files? I guess you can copy them right off the internal hard disks of the recorders.

gnight
boot_me

[daverich]

surely to do 1 bit you need 1 bit converters?


Kind regards

Dave Rich

[Boot_me]

Originally Posted by daverich

surely to do 1 bit you need 1 bit converters?


Kind regards

Dave Rich

Hi Dave,
The Korg MR1 and MR1000 both record at 1 bit, one at 2.something megaherz, the other at 2x that. They have built in hard disks, and output over a usb2.
Korg supplies a program that will down convert it and remove dc offset once its in your pc.

Thats in one of the articles.
There is a bunch of stuff in how people are using them. They sound interesting, but once I have the file... I need to use it


regards
boot_me

[JasonTheron]

I think if the music industry does adopt this standard, the record companies will spend the next twenty years in court to keep it's production out of the consumers' hands (i.e. Suped Disc burners etc).
That might actually be the only way they can continue to make money.

[plush2]

It's an innovative idea but would require a massive turn-over in hardware. The idea of recording to a hardware box, transferring into the computer and then downsampling for mixing seems a bit counterproductive. (no pun intended) It would be cool if it were a straight path from recording to delivery but I can't see that happening in the near future.

[Xenakios]

Originally Posted by JasonTheron

I think if the music industry does adopt this standard, the record companies will spend the next twenty years in court to keep it's production out of the consumers' hands (i.e. Suped Disc burners etc).
That might actually be the only way they can continue to make money.

Well, SACD is already heavily DRM-laden and as far as I know, no solution even exists to play them back on computers. Really useful new technology...NOT.

Of course the analog hole always exists...I'm sure that if the record companies go to the route of releasing new material only on SACD, pirates will simply settle on playing back the material via analog outputs, recording the result digitally, and off to the Net it goes...Again...(This would be in contrast to the current situation where it's possible to bit-perfectly rip the contents of an audio CD digitally and distribute it to the net and/or make physical counterfeit copies...)

I voted for the pizza, coming soon...??

-X

[Boot_me]

Originally Posted by JasonTheron

I think if the music industry does adopt this standard, the record companies will spend the next twenty years in court to keep it's production out of the consumers' hands (i.e. Suped Disc burners etc).
That might actually be the only way they can continue to make money.

I can only think of a few formats where thats been able to be possible, ie. Nikon's proprietary format of RAW. They keep changing the built in equipment identifier, so new camera's files don't work on old software.
I think it is more likely that once these one bit supersampled recordings get into the computers, they will be converted somehow into raw data, then later after any editing is done, perhaps it could be written back out as an original type file. Its all about keeping the data pure.
A lot of software I see written just uses libraries that have been written to read certain file formats. I know that some folks actually write their own.


>Xenakios
>Quote:

>Well, SACD is already heavily DRM-laden and as far as I know, no >solution even exists to play them back on computers. Really >useful new technology...NOT.


Korg gives out a player that also down samples. If the quality is actually the best, I am going to buy one. Supposedly the output when its changed to PCM is at least as good as natively sampling... so why not get one?
Did anyone check out the audio samples? Its interesting that the Cello recorded by the tiny pocket recorder with the AT822 Mic sound to me way better than the big brother which has the full xlr inputs (different mikes). I am still learning the terminology here... but I am not too sure how these machines compare spec wise with something like a emu 1616m... or the Tascam HDP2. I still haven't figured out a lot about recording. Have a lot of reading material I've acquired still left to read.

I do know though, that when they made their recordings I would liked to have heard the native 24/96 Zoom H4 recording... not a mixed down 44.1. I guess they were trying to show the relative quality at the end of a down sample.

Oh yes...
>plush2

>It's an innovative idea but would require a massive turn-over >in hardware.

Which is already going on.

> The idea of recording to a hardware box, >transferring into >the computer and then downsampling for >mixing seems a bit >counterproductive.

Isn't that what a lot of people do now? Hence sample converters? Ditherers?

>It would be cool if it were a straight path from recording to >delivery but I can't see that happening in the near future.

There needs to be a free alternative file type that provides equivalent quality. At that point, libraries for reading and writing of those files become possible.

For $8000 you can get a Copy of Sanoma with a USB key and 8 inputs and outputs (pci card), but it looks like all the in and out on it are probably through specialized optical io. I like the hard disk recorder better.... Meanwhile I'll hope that the Korg software will sample to good format to use with Reaper. And I need to register

regards...

boot_me

PS for a good desription of some of the technology, this is a decently scientific article.
http://www.edn.com/index.asp?layout=...cleid=CA276213

and here is an entire forum on high resolution audio.
http://forums.afterdawn.com/forum_view.cfm/90
m

[daverich]

it seems to me 1-bit recording is an interesting idea but it's still a compromise.

What we need is an audio format with an inherent 2 sample latency.

that way you could actually recreate a square wave.

No matter how quick 1-bit goes it'll still cant say "full power - NOW"

Kind regards

Dave Rich

[Boot_me]

Originally Posted by daverich

it seems to me 1-bit recording is an interesting idea but it's still a compromise.

What we need is an audio format with an inherent 2 sample latency.

that way you could actually recreate a square wave.

No matter how quick 1-bit goes it'll still cant say "full power - NOW"

Kind regards

Dave Rich

HI Dave
After reading that article on how SACD works, I wonder if that is true. I don't know how these things are sampled from the ground up, but if you are sampling 1 bit at 5 + megaherz, As in the Korg M1000 and the standard for the format are set at half that, isn't that effectively what you are talking about? (minus a bunch of math

I don't know how Niquist rates relate to blah blah. But its interesting that they put double sample rate on the box with xlr inputs, and only straight rate on the hand-held. Could this be your square wave? 2 bits times the base rated freq of the format?

best regards
boot_me

[daverich]

Originally Posted by Boot_me

HI Dave
After reading that article on how SACD works, I wonder if that is true. I don't know how these things are sampled from the ground up, but if you are sampling 1 bit at 5 + megaherz, As in the Korg M1000 and the standard for the format are set at half that, isn't that effectively what you are talking about? (minus a bunch of math

I don't know how Niquist rates relate to blah blah. But its interesting that they put double sample rate on the box with xlr inputs, and only straight rate on the hand-held. Could this be your square wave? 2 bits times the base rated freq of the format?

but I could be talking out my arse...
best regards
boot_me

not really.

the 1 bit system basically is +1 or -1.

What you need is a system where the converter can be told that within the next two samples the power of the signal is going from nothing right the way up to full power and stay there.

Kind regards

Dave Rich.

[Boot_me]

Originally Posted by daverich

not really.

the 1 bit system basically is +1 or -1.

What you need is a system where the converter can be told that within the next two samples the power of the signal is going from nothing right the way up to full power and stay there.

Kind regards

Dave Rich.

HI Dave
After reading that article on how SACD works, I wonder if that is true. I don't know how these things are sampled from the ground up, but if you are sampling 1 bit at 5 + megaherz, As in the Korg M1000 and the standard for the format are set at half that, isn't that effectively what you are talking about? (minus a bunch of math

I don't know how Nyquist rates relate to blah blah. But its interesting that they put double sample rate on the box with xlr inputs, and only straight rate on the hand-held. Could this be your square wave? 2 bits times the base rated freq of the format?

but I could be talking out my arse...

My belief albeit a non professional one, is that a digital image of a sound wave is only as good as the circuit that feeds it into the recorder. They make these nice pres that have all this special stuff done to keep the image clean... then it goes into an input to a sampler. If the guy who designed the sampler doesn't know how to make the super clean audio wave hit the sampler correctly, it won't be accurate... impedance matching right?
So the faster you can sample a smaller amount of that wave, the more accurate it would be right? So maximum sound instantly would look like a ramp of 1's right up to the point of that samples peak. In pcm, if I remember right, they sample a whole wave... probably depends on the hardware. Then they sample another. Maybe they do half at time, but I would think that it depends on the chip. But because of that Nyquist rule...I think it would make only the first half of the sample accurate even if it was maxed out with PCM.

So its the old parallel vs. Serial argument.

A parallel to this is that Sata hard drives are faster now than Parallel ATA's. But Sata's are Serial...
time for bed.

I don't think its all hype and drm paranoia. Heck, 24/192 samplers are all over the place. It doesn't get much better than that anyways, sound wise (from what I've read)... as long as its all set up and done right.
Quality is inversely proportional to Money.

best regards
boot_me

[plush2]

Originally Posted by Boot_me

>It's an innovative idea but would require a massive turn-over >in hardware.

Which is already going on.

Just because the idea is out there doesn't mean it's catching on. By massive turnover in hardware I'm referring to the number of people you know who don't yet own an SACD player, soundcard capable of playing back such rates, converters/DAW capable of producing content etc. All of these 'people' represent the massive turnover in hardware that I have yet to see practical evidence of.

Originally Posted by Boot_me

> The idea of recording to a hardware box, >transferring into >the computer and then downsampling for >mixing seems a bit >counterproductive.

Isn't that what a lot of people do now? Hence sample converters? Ditherers?

Not if it can be helped. If someone works at a higher sample rate and bit depth the added recording fidelity is just a part (and arguably a small part) of the advantage gained. The marked sonic advantage is in the mixing and processing at the higher rates with the end product then downsampled/bit reduced for the delivery format.
Because this can't currently be done and also represents another massive turnover in software technology (1 bit capable dsp plugins for example) I don't think the change is in any way imminent.

Originally Posted by Boot_me

Meanwhile I'll hope that the Korg software will sample to good format to use with Reaper. And I need to register

Good to have you here and welcome to the free exchange of ideas. Your comments awoke the skeptic inside me but I admire your forward thinking.

[Boot_me]

Plush2 wrte:
>Good to have you here and welcome to the free exchange of ideas.

Thank you. I love it here so far, I'm so surprised I had never heard of reaper right up until I bought my Edirol. They sell the unit with no software. So I started looking at different DAWs and ended up looking at Traktion (or however its sp) and coming across a Wikipedia link about DAWS. Reaper supports more than any of the others....variety wise. Its all hobby to me, but I will be registering, as I've gotten so much use out of this thing playing and learning its incredible.

>Your comments awoke the skeptic inside me but I admire your >forward thinking.

I'm thinking from a poor guy vs. quality standpoint... and also of the field recorder quality into a sample rate quality. I want the best all around, and from a programmers point of view it always ERks me when there is a file format that remains elusively out of reach of my favorite application

peace
boot_me

[Boot_me]

Here is another link on one bit audio...

http://www.digitalpreservation.gov/f...dd000230.shtml

Has a bunch of info on the technology.

m

[Sanskrit]

I know my response is entirely informed by my emotions, and a teenie bit of experience, but here goes: I don't like SACD, I want it to go away. I have spent years putting together a PCM based studio. The thought of purchasing expensive gear (or waiting for it to get cost effective to do so) is a lifetime I don't have. I continue to watch folks go on about the need for higher and higher sample rates and what's the format of choice: freakin' mp3. I don't forsee, or see, folks clammering for 24 bit PCM discs, or CD's for that matter. SACD feels like (largely) a marketing tool designed to further the bottom line of its developers and an effort to protect the copyrighted fortunes of said developers/companies. I've got no problem protecting intellectual property, but there's got to be another way to do it without requiring or encouraging a total makeover of "the project studio." This comment is NOT meant to slag anyone's purchase decisions. Whatever works for 'ya: get your gear on! I suppose its what I said: an emotional response to the inevitable march of tech, and the loss of my Atari, eight track and beta max tapes. Damn.
Snskrt

[Boot_me]

Originally Posted by Sanskrit

I know my response is entirely informed by my emotions, and a teenie bit of experience, but here goes: I don't like SACD, I want it to go away. I have spent years putting together a PCM based studio. The thought of purchasing expensive gear (or waiting for it to get cost effective to do so) is a lifetime I don't have. I continue to watch folks go on about the need for higher and higher sample rates and what's the format of choice: freakin' mp3. I don't forsee, or see, folks clammering for 24 bit PCM discs, or CD's for that matter. SACD feels like (largely) a marketing tool designed to further the bottom line of its developers and an effort to protect the copyrighted fortunes of said developers/companies. I've got no problem protecting intellectual property, but there's got to be another way to do it without requiring or encouraging a total makeover of "the project studio." This comment is NOT meant to slag anyone's purchase decisions. Whatever works for 'ya: get your gear on! I suppose its what I said: an emotional response to the inevitable march of tech, and the loss of my Atari, eight track and beta max tapes. Damn.
Snskrt

Hi Sanskrit,
My only interest is in being able to edit the base files, that they created SACD with. I've heard SACD has been dead for a while now, meanwhile they are rolling out blue ray, hd dvd, hd radio, etc, and SACD disk count is reaching 4000 (I read somewhere). Its inevitable that higher quality comes in.
I mainly wanted to try to make a place where people could leave links about the file formats so that if Cockos wants to add import or even import/export it would be an easy source of info, and to help those of us considering portable recorders such as the KORG MR1 or MR1000 with info on what we've bought and how we can use it.
As to PCM, I don't like the idea of dithering noise, and if it were possible to record digitally perfect and then convert that into say, a Monkey Audio file (I guess Lossless?) then you would have a way of editing your highly high def recordings.
Its clear to me that no $699 or $1000 recorder is going to take the place of the really good gear.... just like calling a pc box a mastering converter doesn't really make it a mastering converter. The converter may be master grade, but the circuitry that hooks to it has to be that good also. 120 db has to be fed by that quality of electronics and the right room and the right recording techniques.
As a friend of mine said... he's not recording Classical, he's recording Rock. That dictates the quality to some extent I suppose... but if you are using different mediums to get sound into your productions, you might just want lossless audio at some point. I don't know how reaper represents audio once its inside the DAW... true waveforms, pcm code? My friend says it sounds good.
I do think that the portable converters that people listen to files on will get better and cheaper, as well as the algorithms that they save the music in. (My little RCA Pearl is 1 gig and takes micro SD cards, and is max 128 bit mp3. But it also plays and records wave files and audio book files...and cost $43.00)
And I have read a little on the 1 bit technique and see that its based on filtering the signal after its recorded true. I'm thinking maybe a .dsd to monkey audio converter may be the ticket...although I don't know much about Monkey Audio, I read that it is supposed to be lossless... don't know how good or what data form it uses. But some editable true representation of the waveforms.

reguards
boot_me
btw trying to find a dsdiff to ape converter found me another reference to how dsd works.
http://www.digsrex.com/DSD_archival_solution.htm
and this:
http://www.sa-cd.net/faq
And this one has a bunch of good links. Their page is a goldmine of 1 bit documentation:

http://sonicstudio.net/support/knowl....html#nexStage

[toggle]

Boot_me,

A couple of things hung me up here...

Originally Posted by Boot_me

So the faster you can sample a smaller amount of that wave, the more accurate it would be right?

... up until just over 2X your bandwidth.

Originally Posted by Boot_me

So maximum sound instantly would look like a ramp of 1's right up to the point of that samples peak.

I don't think any analog stage is going to provide you with a perfectly square square wave. To have that kind of infinitely fast rise time, you'd need an infinite frequency response.

Originally Posted by Boot_me

In pcm, if I remember right, they sample a whole wave... probably depends on the hardware. Then they sample another. Maybe they do half at time, but I would think that it depends on the chip. But because of that Nyquist rule...I think it would make only the first half of the sample accurate even if it was maxed out with PCM.

I'm not sure what you mean here. There's no way that the converter knows what a whole wave is--it's just sampling the voltage of the signal at regular intervals. As long as the highest frequency present in that signal is lower than 1/2 the frequency of the sampling rate, you're good.

Originally Posted by Boot_me

I want the best all around, and from a programmers point of view it always ERks me when there is a file format that remains elusively out of reach of my favorite application

Originally Posted by Boot_me

My only interest is in being able to edit the base files, that they created SACD with.

I can relate to your quest, but maybe you'll rest easier knowing that DSD material cannot be processed in any way (volume, pan, etc., not to mention EQ and more) without introducing more bits, which in practice means that DSD material is brought down to a more typical sample-rate and bit-depth for processing.

Originally Posted by Boot_me

As to PCM, I don't like the idea of dithering noise, and if it were possible to record digitally perfect and then convert that into say, a Monkey Audio file (I guess Lossless?) then you would have a way of editing your highly high def recordings.

The noise resulting from converting from analog to DSD is pretty substantial itself, though most of it is above the generally accepted threshold of human hearing. The quantization noise in a 24-bit PCM signal is way down there approaching -140dB--well below the self-noise of your mic, pre, and whatever else gets between the sound and the converter...

Monkey's Audio is a lossless compression format. I can't see what advantage it would have over uncompressed PCM, in terms of fidelity.

-Tom

[antiClick]

Originally Posted by daverich

not really.

the 1 bit system basically is +1 or -1.

What you need is a system where the converter can be told that within the next two samples the power of the signal is going from nothing right the way up to full power and stay there.

Kind regards

Dave Rich.

You seem to assume that there exists a full scale power.
In nature full scale power does not exist, we have to always talk about relative scales (dB).

For what I understand a perfect square wave is is impossible in nature because values can't be discrete (to go from 1 to 2 you have to pass on all the infinitum values in the middle), so it also is in DSD.


BTW.... why do you want a perfect square wave? your speakers will never be able to reproduce that

[aldi]

Originally Posted by antiClick

You seem to assume that there exists a full scale power.
In nature full scale power does not exist, we have to always talk about relative scales (dB).

For what I understand a perfect square wave is is impossible in nature because values can't be discrete (to go from 1 to 2 you have to pass on all the infinitum values in the middle), so it also is in DSD.


BTW.... why do you want a perfect square wave? your speakers will never be able to reproduce that

damn right, it's like teleporting: beeing here and then there without passing the miles between. hmm, we should have that and on the enterprise and "the fly" it's already possible.

[daverich]

Originally Posted by antiClick

You seem to assume that there exists a full scale power.
In nature full scale power does not exist, we have to always talk about relative scales (dB).

For what I understand a perfect square wave is is impossible in nature because values can't be discrete (to go from 1 to 2 you have to pass on all the infinitum values in the middle), so it also is in DSD.


BTW.... why do you want a perfect square wave? your speakers will never be able to reproduce that

you look at a square wave from an analog synth on a scope.

It is square. - ok, it might not be square to the atomic level, but it is square.

Digital is nowhere near it - so just fast enough to do that would be good.

Kind regards

Dave Rich

[antiClick]

Originally Posted by daverich

you look at a square wave from an analog synth on a scope.

It is square. - ok, it might not be square to the atomic level, but it is square.

Digital is nowhere near it - so just fast enough to do that would be good.

? I think you are confusing the LFO square pulse that is part of a synth process, with its resulting sound.

But on the other hand, you can clearly see in an osciloscope that an analog square is not perfectly square... no need for an atomic microscope


That's why DSD has a very fast megahrtz clock. Plus since its values are modal instead of scalar, it's like having an infinite samplerate.

Think of it as if it was as a fractal. Infinite per se.


DSD is not a digital representation, but a way to store information digitally that will reconstruct a true analog model (in mostly the same way the analog tape did)


Back on topic,
I don't think DSD will cope the typical PCM industry base. It can't be edited easily, and PCM is very good & fast at that. Maybe maybe someday 1bit technology could help typical upsample/downsample AD converters, but the end result will have to be PCM (not much gain for such a tech effort, if we compare it to straight 192khz PCM).

...we will see, but you can swear that the DAWs will never process DSD

Regards!

[griz lee]

wow. that's a thread.

two things to add:

-- the best recent 1 bit stuff i've heard now uses multibit converters (4 bits wide, at lower rates) to do the D/A conversion.

-- yes, as was said earlier 1 bit causes major DSP headache. To get anywhere even with the best current 1-bit authoring environments, you need to convert to very high rate PCM anyway (well, it's kinda done behind your back in one system) or more openly in the case of Pyramix's DxD format. I know of only a handful of algorithms that work directly on the DSD stream, so I don't expect to see fully 1-bit vintage comp emulations any time soon - just changing the gain of a signal is a surprisingly involved operation in true one bit land.

-- i'm often told by mastering friends that an alarming amount of supposedly '1-bit' material out there was infact sourced on ProTools and recorded to one-bit on the mix output.

I think Pyramix got it right. Doing delta/sigma conversion to 384khz or higher at 24 bit gives excellent performance for processing a DSD recording using the tools which we all know and love largely without modification.

Then treat SACD just as a distribution medium, and convert back. Forget about trying to keep the chain of dsd for highly processed/produced music (at least at this point in history. I suppose it pays to keep an open mind...)


personally, i'll stick to multibit because i really love the sound of 48 kHz for some material.

[Boot_me]

I kind of lost track of this thread, but I was researching dithering schemes and came across this post on Gear Slutz.
As of June 07
> Weiss Saracon updated
>The Weiss Saracon sampling rate converter version 1.5 is >available in two versions, the Saracon and the Saracon-DSD. As >the name implies, the difference between the two versions is the >DSD feature.

>Saracon-Light users are entitled to upgrade to Saracon for free. >Version 1.5 can be downloaded from our website (password >protected).

>Saracon-DSD can do PCM to DSD conversion. DSD to PCM is in >preparation.

>For more information see Weiss :: SARACON
>Contact Daniel Weiss at weiss@weiss.ch

and while I was looking... I came across this.
http://community.sonikmatter.com/for...howtopic=36152

It talks about different dithering techniques, and a comparison of them. They say that basically a bunch of techniques don't compare favorably to the Weiss stuff, but mention at the end, one technique that does (iZotope SRC), and that the creator is licensing "Free".
Wondering if someone should license it and get it into a dither JS Dither for Reaper?

I see that Reaper is in the comparison...

I also see that this has been covered in a different Reaper discussion
http://www.cockos.com/forum/showthre...ighlight=IIRx2

They refer to the final Render, but what about the Dithering? What about Rendering Extreme with PSycho Dither turned on... how would that look?