DSD vs PCM resolution

[Paul R]

You can calculate the rate of a simple conversion that can be accurate. It the sampling rate is 2^n times higher than the PCM sample rate then a simple thermometer code can represent the 2^n levels that are encoded by n bit linear PCM. Some extra margin will be needed to avoid filter ringing and finite word length calculations. (About 3 Gbps for 44/16 PCM). There may be more efficient bit perfect encodings based on noise shaping or other encoding techniques. I am not saying that this data rate is actually necessary for a lossless conversion. I don't know the lower bound.

 

 

Well:

 

1. You cannot convert 16/44.1 -> 24/96 -> 16/44.1 and get a perfect reconstruction. Unless you did the up sampling by sample and hold, essentially duplicating each sample 51,900 times. If you use interpolation, no way you can reconvert to 16/44.1 with perfect accuracy. So no - you cannot convert 44.1 -> 96 -> 44.1 without error and loss.

 

2. Same applies to DSD, if you do the conversion in such as way as to preserve the original samples, you an convert back, but it would be about as useless in audiophile terms as converting 16/44.1 to 24/96 with sample and hold.

 

3. Again, you are assuming that a conversion has to be reversible to be "lossless" and that is not a true assumption. Even mathematically. Neither the upsample to 96K or the conversion to DSD looses any data - in other words they are perfectly lossless.(*sigh*)

 

Okay - I am out of it. You guys go ahead and believe what you want. Listen to those very lossy up sampled PCM music files and I'll listen to the lossless DSD files I enjoy so much. ]8)

 

This is the internet after all - it is not a emergency if someone else says something wrong. :)

 

P.S. I am still willing to be convinced otherwise, if someone can show me an authoritative reference that defines lossless the way you are assuming it to be - without a reference to compression theory. Just pure format conversion. No wikipedia please.

[Hiro]

Well:

 

1. You cannot convert 16/44.1 -> 24/96 -> 16/44.1 and get a perfect reconstruction. Unless you did the up sampling by sample and hold, essentially duplicating each sample 51,900 times. If you use interpolation, no way you can reconvert to 16/44.1 with perfect accuracy. So no - you cannot convert 44.1 -> 96 -> 44.1 without error and loss.

 

2. Same applies to DSD, if you do the conversion in such as way as to preserve the original samples, you an convert back, but it would be about as useless in audiophile terms as converting 16/44.1 to 24/96 with sample and hold.

 

3. Again, you are assuming that a conversion has to be reversible to be "lossless" and that is not a true assumption. Even mathematically. Neither the upsample to 96K or the conversion to DSD looses any data - in other words they are perfectly lossless.(*sigh*)

 

Okay - I am out of it. You guys go ahead and believe what you want. Listen to those very lossy up sampled PCM music files and I'll listen to the lossless DSD files I enjoy so much. ]8)

 

This is the internet after all - it is not a emergency if someone else says something wrong. :)

 

P.S. I am still willing to be convinced otherwise, if someone can show me an authoritative reference that defines lossless the way you are assuming it to be - without a reference to compression theory. Just pure format conversion. No wikipedia please.

 

We also have to remember that PCM carries a lot of redundant information from sample to sample to describe an analog wave. Something like MQA packs PCM data into a space that's a fraction of the original PCM file.

[Tony Lauck]

Tony,

 

May be lost of translation (sorry for my English).

 

Here 4 points:

 

 

1. When we resample, signal always have energy in passband ("...one half of the lower of the old and new sampling rates...").

 

2. Analog filters used in ADC and DAC for any sample rate for both PCM and DSD.

 

Otherwise mirrored useful spectrum (all that upper sample rate /2) can be shifted to 0 ... sample rate/2.

For ADC without filtration will periodically mirrored parts of all spectrum [0 ... infinity] due ambiguity of measurements for frequencies above [sample rate/2].

 

 

3. Filters used for mixing / postproduction. Often it is real-time IIR filters what distort phase response.

 

4. We can decrease ringing with lesser steepness. But we can't say about filter with "zero ringing".

 

 

 

1) Could you comment it in link with your words ("...input signal has no energy...")?

 

2) Could you illustrate (spectrum, scheme pictures or other way) what you said above?

 

 

All of my discussions concerned digital filters, as used in computer software or in the digital portion of DAC hardware.

 

My comment about filter ringing are well known. They can be demonstrated by using an audio editor that has a flexible set of resampling filters. An example would be iZotope RX4 advanced, which I use for audio restoration work and general PCM format conversions. I suggest getting a good audio editor and doing these experiments yourself. You can adjust the graphics to suit your eyesight. You can use test tones or music. You can even listen and hear the differences involved. One of the things you will learn is how filters used for downsampling (on recording) and upsampling (on playback) interact with each other and how the choice of best playback filter will depend on the filter used in recording.

 

Another way of understanding this situation is theoretical. A band limited signal can be interpolated (and hence upsampled) using the sinc filter. This will provide perfect reconstruction of the band limited signal, despite the fact that the sinc function rings infinitely, both pre and post when excited by an impulse. Worse, the filter is actually unstable if presented with certain input signals (e.g. 44/16 files) that do not conform to the requirements of the sampling Theorem. However, if the input signal was already band limited to well less than Fs/2 by an Apodizing filter, then the sinc filter will add no ringing at all.

 

RX4 has a choice of minimum phase filters (with no preringing) and linear phase filters (which have preringing). I usually use minimum phase filters, which are FIR approximations to IIR filters because on most program material they sound better and they have a limited downside in terms of sonic damage, at least to my ears.

 

There are filters that have zero ringing. A simple first order RC filter with 6 dB per octave roll off does not ring. The best possible filter (steepest possible slope) is a Gaussian filter. From the formula in the linked article you can work out the possibilities. https://en.wikipedia.org/wiki/Gaussian_filter

[Tony Lauck]

Well:

 

1. You cannot convert 16/44.1 -> 24/96 -> 16/44.1 and get a perfect reconstruction. Unless you did the up sampling by sample and hold, essentially duplicating each sample 51,900 times. If you use interpolation, no way you can reconvert to 16/44.1 with perfect accuracy. So no - you cannot convert 44.1 -> 96 -> 44.1 without error and loss.

 

2. Same applies to DSD, if you do the conversion in such as way as to preserve the original samples, you an convert back, but it would be about as useless in audiophile terms as converting 16/44.1 to 24/96 with sample and hold.

 

3. Again, you are assuming that a conversion has to be reversible to be "lossless" and that is not a true assumption. Even mathematically. Neither the upsample to 96K or the conversion to DSD looses any data - in other words they are perfectly lossless.(*sigh*)

 

Okay - I am out of it. You guys go ahead and believe what you want. Listen to those very lossy up sampled PCM music files and I'll listen to the lossless DSD files I enjoy so much. ]8)

 

This is the internet after all - it is not a emergency if someone else says something wrong. :)

 

P.S. I am still willing to be convinced otherwise, if someone can show me an authoritative reference that defines lossless the way you are assuming it to be - without a reference to compression theory. Just pure format conversion. No wikipedia please.

 

1. This is not true. You can get a very good conversion from 44.1 to 96 and back to 44.1 providing that you use the proper software with the proper settings and providing that the source file was already apodized. The limitation will be due to different dither noise. I suggest using iZotope RX4 advanced and running some tests yourself to confirm this. If you started with 44.1/24 and went to 96/32 and back to 44.1/24 you would still have much more than 16 bits of accuracy. However, if you were to round the two 24 bit files to 16 bits you might see a few samples where the rounding was different.

 

2. You can not do this with DSD as the middle format, you can't even come close.

 

3. If you can not get the original back you have lost data. Period. Mathematically you have used a many to one function, which is not invertible. If you were to consider the input file as your data then if you didn't get back the identical data you would have lost data. The reason for going lossless is not because it may be sonically necessary, but because if you get back all the bits you started with then there is nothing to argue about when it comes to the process changing the sound quality.

 

DSD sounds better to many people than PCM (me included) because DSD does not require an antialiasing filter. However, DSD sounds worse than PCM to many other people because the modulators have artifacts and the high frequency noise may interact with downstream equipment.

[Hiro]

DSD sounds better to many people than PCM (me included) because DSD does not require an antialiasing filter. However, DSD sounds worse than PCM to many other people because the modulators have artifacts and the high frequency noise may interact with downstream equipment.

 

As we move to higher DSD sampling rates, the high frequency noise becomes a non-issue. Have a look at these measurements of Miska's DSC1 DAC.

 

Squarewaves from DACs - Blogs - Computer Audiophile

[audiventory]

All of my discussions concerned digital filters, as used in computer software or in the digital portion of DAC hardware.

 

My comment about filter ringing are well known. They can be demonstrated by using an audio editor that has a flexible set of resampling filters. An example would be iZotope RX4 advanced, which I use for audio restoration work and general PCM format conversions. I suggest getting a good audio editor and doing these experiments yourself. You can adjust the graphics to suit your eyesight. You can use test tones or music. You can even listen and hear the differences involved. One of the things you will learn is how filters used for downsampling (on recording) and upsampling (on playback) interact with each other and how the choice of best playback filter will depend on the filter used in recording.

 

Another way of understanding this situation is theoretical. A band limited signal can be interpolated (and hence upsampled) using the sinc filter. This will provide perfect reconstruction of the band limited signal, despite the fact that the sinc function rings infinitely, both pre and post when excited by an impulse. Worse, the filter is actually unstable if presented with certain input signals (e.g. 44/16 files) that do not conform to the requirements of the sampling Theorem. However, if the input signal was already band limited to well less than Fs/2 by an Apodizing filter, then the sinc filter will add no ringing at all.

 

RX4 has a choice of minimum phase filters (with no preringing) and linear phase filters (which have preringing). I usually use minimum phase filters, which are FIR approximations to IIR filters because on most program material they sound better and they have a limited downside in terms of sonic damage, at least to my ears.

 

There are filters that have zero ringing. A simple first order RC filter with 6 dB per octave roll off does not ring. The best possible filter (steepest possible slope) is a Gaussian filter. From the formula in the linked article you can work out the possibilities. https://en.wikipedia.org/wiki/Gaussian_filter

 

 

I understand you, Tony.

 

But all filters that you called (including Gaussian) has pre-action and post action. Anyway we get smoothing of signal (distortions).

 

 

 

6 dB per octave is not suitable even for lo-fi filter for DSD512: 512=2^9 -> 9*6db/oct=54 dB

 

For studio quality filter here must be 170 ... 200 dB in 20...22 kHz. Otherwise we catch noise, that should not catch.

 

In my opinion, better way suppress DSD noise from and above point of minimal noise floor value (about 20 kHz). It help keep useful information only.

 

 

 

However, if the input signal was already band limited to well less than Fs/2 by an Apodizing filter, then the sinc filter will add no ringing at all.

 

If I correct understand you, it is next scheme:

 

input signal -> min-phase-filter (w/o pre-ringing) -> linear-phase filter (with pre-ringing) -> output signal

 

However in this scheme min-phase filter have double energy post-ringing (comparing with linear phase filter) and we add pre- and post-ringing/action by liner-phase filter.

 

I don’t see necessity apply linear phase filter (2nd stage) here. It give additional distortions and additional time of calculations.

 

Simpler way is optimizing min-phase filter (if goal is pre-ringing elimination) to better compromise: ringing-suppression-length-amplitude response flatness-phase distortions.

[magnum innominandum]

Mes Ami Paul,

 

Perhaps you should re-read a few articles - PCM to DSD conversion is perfectly lossless. No audio or timing information whatsoever is lost. It is a conversion, -> not <- a compression algorithm.

 

If this was the case you could do the following:

 

1) Start with any music file in 16Bit/44.1kHz extracted from CD.

2) Convert to DSD64 using any conversion software you care to use.

3) Convert back from DSD64 to 16Bit/44.1kHz.

4) Run the original source file and the result of the double conversion through Audio DiffMaker:

Audio DiffMaker

 

If the conversion is completely lossless the two files will be identical.

 

If they are not, you will able to analyse (Listen, FFT, virtual oscilloscope etc. et all) precisely what the differences are.

 

If you find the results identical - please post the results here, including the source file and what converter you used with which settings. If you find them different, you may also wish to post the results and a retraction.

 

Salud M.I.

[magnum innominandum]

Bonsoir,

 

But at the end of the day - you gotta listen to music. You can't enjoy music by looking at bits and graphs. And the reason I listen to DSD is simple. I can listen to DSD without a DAC. And I do. Regardless of the technical merits of FSD vs PCM, I reckon that the physical implementation of the DAC is the biggest determinant of sound quality. And I find that DSD, played through a simple LPF, is better than any DAC. The sound has a holographic (analog) character that i don't experience from any DAC.

 

[ATTACH=CONFIG]20170[/ATTACH]

 

I did try a very similar "no-chip" DAC (DIYINHK USB Board with isolator and external clock) after it was talked up so much in some babillards. I used 3rd order passive filter at 100kHz.

 

Using Foobar2k's DSD converter I tried this at the different supported DSD rates. Even at DSD256 I felt the original Pass converter with PCM63 DAC Chip's did a much better job of sounding "analogue" or "like the best of vinyl minus clicks, pops, noise and tracking distortion". At slower speed the advantage of the Pass converter increased, I felt CD->DSD64 by comparison produced a hilarious dummy sound, caricature de la musique de merde.

 

But the killer was listening to actual DSD downloads. I found that many DSD files I played via this kind of "chipless DSD DAC" had very audible distortion and background sounds I would call "birdies" (based on my Ham radio days). Merde de merde.

 

Funny thing, DSD converted to 88.2kHz PCM and played via Pass D-1 sounds okay, but not as a good CD.

 

Salud M.I.

[Jay-dub]

I do not believe PCM to DSD conversions can be exact, at least not 44/16 to DSD64. I don't believe this conversion can even pass the following, somewhat relaxed test where 44/16 is converted to DSD and back to 44/16:

 

2. The decoder must perform the same function digitally as would an analog DSD decoder, namely some kind of low pass filtering and the result of this is then converted back to 44/16.

 

 

Please note that the essence of this test is the almost bit perfect comparison. Mathematically it consists of using the L-infinity norm. The usual measurement of noise and distortion uses the L2 norm, a.k.a. RMS. The RMS norm implies equality only if it is a true zero. Impressive numbers of dB down RMS does not necessarily say anything about the accuracy unless the averaging period is also specified.

 

Do you have in mind some sort of nonlinear modulation artifacts that you would expect any DSD modulator add, causing it to fail, Tony? This is clearly a well thought out test, though I'm still thinking about the implications of the parts quoted here. I agree it is possible for 16/44.1 -> 24/96 -> 16/44.1 to pass this test as described here. However, I haven't played around with DSD modulators or explored their theory enough to get why you think they'll fail this test.

 

In the PCM case, you have to be very careful to line up the time and level of the input and output, but if you use truncation/rounding instead of dither, then you pass the test if the double rate conversion at high resolution adds no errors greater than 1.5 LSB at any sample. Filters with flat frequency response in the passband and stopband attenuation are definitely possible to the required level of accuracy, so the main difference between input and high resolution output at 44.1 will result from filtering and/or aliasing of the quantization noise above 20kHz in the input. This is sufficiently low energy that even if you rely on a statistical model for the noise rather than a worst-case analysis for specific dithers and filters, you will still get a high probability of success in 75 min of signal.

 

The output of a DSD modulator can be analyzed as original signal plus quantization noise, where the noise cannot be uncorrelated with the signal as in the PCM case but still is generally quantified without reference to specific signals. It normally has energy in the 0-22 kHz band considerably lower than the RBCD quantization noise has in the 20 kHz -22 kHz band. Analog lowpass filters won't be flat enough for perfect reconstruction, but digital filters matched to specific modulators shouldn't have a problem. Where do you expect DSD to fail?

[Hiro]

Using Foobar2k's DSD converter I tried this at the different supported DSD rates. Even at DSD256 I felt the original Pass converter with PCM63 DAC Chip's

 

Just checked that this PCM chip has THD+N at rbcd levels, or worse.

 

Here's another marvel of PCM technology, BB PCM1700, with its out-of-band noise.

 

[Hiro]

DSD sounds better to many people than PCM (me included) because DSD does not require an antialiasing filter. However, DSD sounds worse than PCM to many other people because the modulators have artifacts and the high frequency noise may interact with downstream equipment.

 

I'm afraid that some people simply got used to typical PCM artifacts resulting from brickwall filtering, decimation filtering, aliasing and ringing, and when they hear something different, they just opt for what they are more familiar with.

[Paul R]

Mes Ami Paul,

 

 

 

If this was the case you could do the following:

 

1) Start with any music file in 16Bit/44.1kHz extracted from CD.

2) Convert to DSD64 using any conversion software you care to use.

3) Convert back from DSD64 to 16Bit/44.1kHz.

4) Run the original source file and the result of the double conversion through Audio DiffMaker:

Audio DiffMaker

 

If the conversion is completely lossless the two files will be identical.

 

If they are not, you will able to analyse (Listen, FFT, virtual oscilloscope etc. et all) precisely what the differences are.

 

If you find the results identical - please post the results here, including the source file and what converter you used with which settings. If you find them different, you may also wish to post the results and a retraction.

 

Salud M.I.

 

Again, I contend this is a mistaken assumption. That is not what lossless means, except in terms of compression algorithms.

 

So, in your opinion:

 

1. Is a file that is up sampled from 16/44.1 to 24/96 lossless or not?

 

2. Exactly what information is lost in converting PCM to DSD?

 

-Paul

[Miska]

1. This is not true. You can get a very good conversion from 44.1 to 96 and back to 44.1 providing that you use the proper software with the proper settings and providing that the source file was already apodized. The limitation will be due to different dither noise. I suggest using iZotope RX4 advanced and running some tests yourself to confirm this. If you started with 44.1/24 and went to 96/32 and back to 44.1/24 you would still have much more than 16 bits of accuracy. However, if you were to round the two 24 bit files to 16 bits you might see a few samples where the rounding was different.

 

2. You can not do this with DSD as the middle format, you can't even come close.

 

Wrong, since you can get way over 24-bit worth of accuracy within audio band with DSD.

 

However, for playback this is irrelevant because only thing that matters is what comes out of DAC output. Since practically all DACs are oversampling delta-sigma type, the question is which way gives best output performance.

 

3. If you can not get the original back you have lost data. Period. Mathematically you have used a many to one function, which is not invertible. If you were to consider the input file as your data then if you didn't get back the identical data you would have lost data. The reason for going lossless is not because it may be sonically necessary, but because if you get back all the bits you started with then there is nothing to argue about when it comes to the process changing the sound quality.

 

As long as the error bounds of the conversion far exceed precision of the input, you can go back and forth without loss.

 

DSD sounds better to many people than PCM (me included) because DSD does not require an antialiasing filter. However, DSD sounds worse than PCM to many other people because the modulators have artifacts and the high frequency noise may interact with downstream equipment.

 

Good modulators don't have artifacts, bad ones do.

[Tony Lauck]

I'm afraid that some people simply got used to typical PCM artifacts resulting from brickwall filtering, decimation filtering, aliasing and ringing, and when they hear something different, they just opt for what they are more familiar with.

 

My standard was, and still is, a live microphone feed vs. something recorded and then played back. I never "got used to" any kind of distortion, be it tape hiss, tape saturation, vinyl distortion due to geometry errors or mis tracking, FM radio transmitter/receiver distortion (including with a Marantz 10B), not to mention the myriad of complex distortions created by the CD process, especially early versions thereof.

 

I don't listen to amplified music, except for jazz vocals and then I do not judge the sound quality based on the vocals, just the instruments. My belief is that people who do not listen to acoustic music are not qualified to judge music the way that I do. They can enjoy their music if they like it. I will enjoy mine. It would be better if there were separate forums for discussing audio reproduction, segregated according to acoustic vs. non-acoustic. Many forum discussions have gone off the rails, until it turns out that the factions are listening to completely different music and judging what they here by completely different standards.

[Miska]

The output of a DSD modulator can be analyzed as original signal plus quantization noise, where the noise cannot be uncorrelated with the signal as in the PCM case but still is generally quantified without reference to specific signals. It normally has energy in the 0-22 kHz band considerably lower than the RBCD quantization noise has in the 20 kHz -22 kHz band. Analog lowpass filters won't be flat enough for perfect reconstruction, but digital filters matched to specific modulators shouldn't have a problem. Where do you expect DSD to fail?

 

There are ways to enforce noise decorrelation...

[audiventory]

Good modulators don't have artifacts, bad ones do.

 

Good modulators have artefacts at proper level. Bad ones can be worse than CD sometime

[Tony Lauck]

Wrong, since you can get way over 24-bit worth of accuracy within audio band with DSD.

 

However, for playback this is irrelevant because only thing that matters is what comes out of DAC output. Since practically all DACs are oversampling delta-sigma type, the question is which way gives best output performance.

 

 

 

As long as the error bounds of the conversion far exceed precision of the input, you can go back and forth without loss.

 

 

 

Good modulators don't have artifacts, bad ones do.

 

Miksa, you keep saying this, but I don't believe there is any evidence to this effect. I don't believe you can even get consistent 16 bit accuracy from DC to 20 kHz with DSD64, once you use the L-infinity norm rather than the L2 (RMS) cover-up of transient errors.

 

Is there some way to do the comparison tests that I suggest with your good modulators? I am talking about the worst case errors in a low pass filtered output. You can select the low pass filters you want. It would be nice if there were some way to input and output to your modulators via a Python program.

[Miska]

Miksa, you keep saying this, but I don't believe there is any evidence to this effect. I don't believe you can even get consistent 16 bit accuracy from DC to 20 kHz with DSD64, once you use the L-infinity norm rather than the L2 (RMS) cover-up of transient errors.

 

Is there some way to do the comparison tests that I suggest with your good modulators? I am talking about the worst case errors in a low pass filtered output. You can select the low pass filters you want. It would be nice if there were some way to input and output to your modulators via a Python program.

 

Which one of these two has gone 44.1/16 -> 96/24 -> 44.1/16 and which one 44.1/16 -> DSD64 -> 176.4/32 -> 44.1/16 and which one 44.1/16 -> DSD512 -> 1411.2/32 -> 44.1/16?

[Miska]

I did try a very similar "no-chip" DAC (DIYINHK USB Board with isolator and external clock) after it was talked up so much in some babillards. I used 3rd order passive filter at 100kHz.

 

That's pretty bad way and you are not conforming to the minimum requirements of DSD specification...

 

Using Foobar2k's DSD converter I tried this at the different supported DSD rates. Even at DSD256 I felt the original Pass converter with PCM63 DAC Chip's did a much better job of sounding "analogue" or "like the best of vinyl minus clicks, pops, noise and tracking distortion". At slower speed the advantage of the Pass converter increased, I felt CD->DSD64 by comparison produced a hilarious dummy sound, caricature de la musique de merde.

 

So you intentionally compared to a flawed implementation and then say that it doesn't meet your performance expectations... Oh well.

[Miska]

Just checked that this PCM chip has THD+N at rbcd levels, or worse.

 

Here's another marvel of PCM technology, BB PCM1700, with its out-of-band noise.

 

[ATTACH=CONFIG]20184[/ATTACH]

 

Note that IIRC, this also has a 3rd order reconstruction filter with fc of 25 kHz. So not really suitable for hires content. If you keep the same 352.8k DAC chip conversion rate as is common still these days but push the reconstruction filter for hires rates to fc 100 kHz, then it becomes much worse.

[Miska]

If this was the case you could do the following:

 

1) Start with any music file in 16Bit/44.1kHz extracted from CD.

2) Convert to DSD64 using any conversion software you care to use.

3) Convert back from DSD64 to 16Bit/44.1kHz.

4) Run the original source file and the result of the double conversion through Audio DiffMaker:

Audio DiffMaker

 

If the conversion is completely lossless the two files will be identical.

 

If they are not, you will able to analyse (Listen, FFT, virtual oscilloscope etc. et all) precisely what the differences are.

 

If you find the results identical - please post the results here, including the source file and what converter you used with which settings. If you find them different, you may also wish to post the results and a retraction.

 

Now do this yourself while capturing an actual DAC output instead of playing with digital domain, and post your results here.

[audiventory]

Which one of these two has gone 44.1/16 -> 96/24 -> 44.1/16 and which one 44.1/16 -> DSD64 -> 176.4/32 -> 44.1/16 and which one 44.1/16 -> DSD512 -> 1411.2/32 -> 44.1/16?

[ATTACH=CONFIG]20187[/ATTACH]

[ATTACH=CONFIG]20188[/ATTACH]

[ATTACH=CONFIG]20189[/ATTACH]

 

I check this via sweep sines. It give more full picture. Due for different frequencies can be different distortions.

 

Though possibly for 16/44 artefacts under noise floor for all frequencies.

[Miska]

There are filters that have zero ringing. A simple first order RC filter with 6 dB per octave roll off does not ring.

 

This is why DSD is so great, first order filter is enough as anti-aliasing filter, while with typical PCM rates you end up requiring much higher order filter.

[Miska]

I check this via sweep sines. It give more full picture. Due for different frequencies can be different distortions.

 

Though possibly for 16/44 artefacts under noise floor for all frequencies.

 

It doesn't change the result...

 

There are no distortions and noise floor even with DSD64 is way much lower than 16-bit PCM resolution. With DSD512 noise floor is way below 24-bit PCM resolution too, so even with 44.1/24 test

[Tony Lauck]

Which one of these two has gone 44.1/16 -> 96/24 -> 44.1/16 and which one 44.1/16 -> DSD64 -> 176.4/32 -> 44.1/16 and which one 44.1/16 -> DSD512 -> 1411.2/32 -> 44.1/16?

[ATTACH=CONFIG]20187[/ATTACH]

[ATTACH=CONFIG]20188[/ATTACH]

[ATTACH=CONFIG]20189[/ATTACH]

 

No spectrum plots or spectrograph plots, please. I am interested only in looking at the time domain performance for this exercise. That is the raw data. I am not interested in statistical analysis, other than my own. I've seen way too many engineering technical presentations where statistics were used to prove a point that was later shown to be false.

 

One other problem I have with these plots. There is no way to calibrate the depth of null without knowing the FFT gain. So, for example, at what level does 16 bit dither noise show up in these plots?