USB dac that shows bit depth on display?

[Bamyasi]

Make yourself a bar graph dB display with peak hold

Or play the file in Audition and set the dB display for peak hold

A bit depth display will not tell you what the song actually is

Peak dynamic range will tell you exactly what the maximum number of bit were used for the song

Saving a song as a 24 or 32 bit file does not tell you the number of bits used for the song itself or how it was originally recorded

A bit depth readout is worthless for telling you what you are actually playing but peak dynamic range would (sort of)

even dynamic range can be tweaked during conversion so you would never know what the original recording actually was

You are digging for fools gold

Please stop your attempts to hijack this thread. No one here discusses the merits of HiRes recordings (or lack of thereof). This was a purely technical discussion about digital formats used in USB audio and the OP question was how to determine bit depth of a USB audio stream. This topic has nothing to do with DR of a recording or with the music in general. Maybe he was going to listening to some synthesized test tones. Please take your discussion of dynamic range in "songs" to another thread.

 

[Bullwinkle J Moose]

Please stop your attempts to hijack this thread. No one here discusses the merits of HiRes recordings (or lack of thereof). This was a purely technical discussion about digital formats used in USB audio and the OP question was how to determine bit depth of a USB audio stream. This topic has nothing to do with DR of a recording or with the music in general. Maybe he was going to listening to some synthesized test tones. Please take your discussion of dynamic range in "songs" to another thread.

and you have your answer....
You cannot determine "actual" bit depth of an audio stream

You may limit the stream to 16 / 24 and 32 bits in Windows or place your audio file in a 16 / 24 or 32 bit container, but you will never know what the content was originally with a bit depth display unless you calculate the peak bit depth from a maximum dB value which your level meter can give you

This is a fact

Nobody is hijacking this thread

I am simply giving you the answer and you don't want to hear it

The DAC will take whatever bit depth you throw at it up to the limit of it's capability, but it will not tell you whether you are sending a 4 bit file or a 20 bit file within a 32 bit container (unless you try that VU metering tip)

You don't seem to get what it is that you don't seem to get

 

[Phorize]

Ok! Then I understand why its rarely implemented, and if not 100% accurate, yes.

Why I wonder these things is for alternatives instead of using a PC. I really like the setup tho.

But there is several products/streamers with USB output. Raspberry Pi is a very cheap alternative, and the Pro-Ject Stream Box S2 Ultra for 800 euros. Looks like it is a Respberry Pi. But using these streamers with no interface seams risky, very risky. How do you know what the output is?

They have an interface via a web gui
You have granular control over bitdepth for some applications using common software like moode on these devices.

 

[Phorize]

What risk do you think there is?

The PTSD induced by not knowing the bit depth

 

[Moderate Dionysianism]

Apparently, the Pro-Ject DAC Box / Pre Box RS2 have this feature:

But apart from this, it seems they only have drawbacks - the DAC costs almost twice as much as the RME, the specs are nothing to write home about apparently, and one year after the launch I can't seem to find a real-life picture of it.

 

[KSTR]

You cannot determine "actual" bit depth of an audio stream

Of course you can with a bit of number-crunching (histogramm binning of sample values), the more data you have (several seconds of "normal" music) the higher the reliability of the bit depth detection.

 

[Bullwinkle J Moose]

Of course you can with a bit of number-crunching (histogramm binning of sample values), the more data you have (several seconds of "normal" music) the higher the reliability of the bit depth detection.

and when exactly would this relevant information be displayed on the DAC ?

After the song has ended, or will you buffer and process the entire song in the DAC to get that info before the song begins ?

and will it be peak, or average, or something else ?

Will it be what the song was originally, or what you converted it to ?

Does he want to accurately display 4 bits if he plays a 4 bit file in a 24 bit container, or does he want to display 24 bits for the 4 bit file ?

It will still sound like a 4 bit file if you convert the song to 24 bits, but how will you accurately determine if it is still a 4 bit file
(now in a 24 bit container) ?

When not using wasapi, does he want to see that Windows is set to 32 bits when he plays his 4 bit file in a 24 bit container, or does he want to see 24 bits displayed ?

The original poster has not specified the information required to make such a determination

There is another way to determine the bit depth that is just as accurate as anything I've heard so far......

Just stick a phoney display on the front of your DAC that says 33 1/3 bits and call it an upgrade

I'm sure your record loving friends can easily hear the difference

 

[mansr]

Most USB DACs support only a single sample size, be it 16, 24, or 32 bits. The OS drivers or audio player will then pad or truncate the content being played as necessary. To display anything meaningful, the DAC would have to examine the data received and check for zero padding, which apparently the RME does.

 

[Hayabusa]

"I have a TEAC NT-503 connected to my Windows PC via USB. I have installed the TEAC drivers (I thinks ASIO). TEAC provides their own playback software. When I use the TEAC software, both the software and the DAC report the correct bit depth and sample rate; e.g..: 16/44.1 or 24/96 "

This was said in https://www.audiosciencereview.com/...-audio-device-sample-rate-and-bit-depth.9587/

 

[KSTR]

Most USB DACs support only a single sample size, be it 16, 24, or 32 bits. The OS drivers or audio player will then pad or truncate the content being played as necessary. To display anything meaningful, the DAC would have to examine the data received and check for zero padding, which apparently the RME does.

Ahm, no. The RME relies on the specific clever bit depth test signals. It constantly analyzes any incoming digital stream and looks for the pattern.

To my knowledge, OS drivers for USB are either 16bit or 32bit on the device level and 32 or 64 bit on the application level. For me, the 32 bit test files pass on Windows (with ASIO or class-compliant driver) either as 24 bit (eg Foobar) or 32bit (eg Reaper), depending on the coding model (Foobar is 32bit floats internally, so no more than 24bits integer precision). On Linux (Mint), with ALSA also I get 32 bits passed from the application (eg Dead Beef player).

 

[mansr]

Ahm, no. The RME relies on the specific clever bit depth test signals. It constantly analyzes any incoming digital stream and looks for the pattern.

Are you saying that display is only valid when playing the special test sequences, not in normal operation? Either way, it looks at the data received.

To my knowledge, OS drivers for USB are either 16bit or 32bit on the device level and 32 or 64 bit on the application level.

The standard USB audio class interface allows sample sizes of 8, 16, 24, or 32 bits with any (smaller) number of effective bits. It is possible for a DAC to advertise, for instance, 18-bit samples in a 24-bit container. In practice, all I've ever seen is 16, 24, and 32 bits for both sample size and effective resolution.

DACs with non-standard interface/drivers can obviously do whatever they want.

 

[KSTR]

Are you saying that display is only valid when playing the special test sequences, not in normal operation?

Exactly. Sequence attached. It's the same pattern for any sample rate (just change the fields in the header).

 

[danadam]

Are you saying that display is only valid when playing the special test sequences, not in normal operation?

For "Bit Test" yes. It looks like this (from the manual):

AFAIK, in case of USB input this is the only option. For SPIDF there is the "State Overview" mentioned earlier:

 

[Tks]

There is none that will report USB bit-depth itself.

 

[Bullwinkle J Moose]

Well the popcorn is ready, so it's time to relax and get another opinion

Hey Amir,

I believe the original poster would like to display the true bit depth of an audio file on a USB DAC

How would you display the true bit depth of a test file recorded @ 20 bits, then converted to 16 bits and normalized to -6dB, then converted to a 24 bit file and played on a Windows box first set to 32 bits, then using Wasapi?

Tell us how to get an accurate bit depth reading from that (on a USB DAC) please

We can make more popcorn if necessary

 

[bennetng]

Of course you can with a bit of number-crunching (histogramm binning of sample values), the more data you have (several seconds of "normal" music) the higher the reliability of the bit depth detection.

I made a tool using a somewhat similar concept. For example, it showed bit discontinuity in GoldenOne's MQA test file (MQA Encoded\882khz.flac). Notice that bit 9-11 are missing.

WAVE_FORMAT_EXTENSIBLE
00:05:00.8850793 = 26538064 samples / 2-ch @ 44100Hz
24-bit fixed point
Bit    Count         Percent
0    177      0.0006669665
1    164      0.0006179803
2    327       0.001232192
3    678       0.002554821
4    1240      0.004672534
5    2564      0.009661594
6    5079       0.01913855
7    10162      0.03829217
8    19959      0.07520895
12    40676       0.1532742
13    81581       0.3074113
14    163500      0.6160962
15    325038       1.224799
16    651069        2.45334
17    1294683      4.878589
18    2552664      9.618878
19    4805265      18.10707
20    7571616      28.53115
21    6542983      24.65509
22    1954869      7.366283
23    513770       1.935974
BitSort end

Another file (Tidal HiFi version\882khz.flac) with bit 1-4 missing

00:05:00.8850793 = 26538064 samples / 2-ch @ 44100Hz
16-bit fixed point
Bit    Count         Percent
0    40350       0.1520458
5    81416       0.3067895
6    163603      0.6164843
7    325056       1.224867
8    651580       2.455266
9    1294673      4.878551
10    2553814      9.623212
11    4809118      18.12159
12    7583667      28.57656
13    6560919      24.72267
14    1959361       7.38321
15    514507       1.938751
BitSort end

People who are interested can find the download link in the post below:
https://www.audiosciencereview.com/...he-obsession-with-dr-meters.11297/post-649576

Notice the date, it is earlier than GoldenOne's post and that means it is not intended to detect MQA files, just a coincidence. The readme html file bundled with the tool explained what it can do, and the limitations. Of course, it is a piece of software, so it can't probe anything related to hardware as well.

 

[julian_hughes]

Hi ASR members!

I'am looking for a USB dac that shows bit depth on its display.
I see sample rate is more common, but I'am most interested in bit depth from different sources.

Thanks in advance!

/ Karl

No need. Just assess it by ear.

 

[danadam]

How would you display the true bit depth of a test file recorded @ 20 bits, then converted to 16 bits and normalized to -6dB, then converted to a 24 bit file and played on a Windows box first set to 32 bits, then using Wasapi?

If it was me, and assuming the normalization was a bit more (less?) than -6 dBFS, let's say -6.03 dBFS, then I'd like:

• 32 / 16 / 15 - if that conversion to 24 bit was just padding with 0
• 32 / 24 / 23 - if that conversion to 24 bit was with added dither

In case DAC accepts only one "wire" format, the first number can be dropped.

 

[KSTR]

How would you display the true bit depth of a test file recorded @ 20 bits, then converted to 16 bits and normalized to -6dB, then converted to a 24 bit file and played on a Windows box first set to 32 bits, then using Wasapi?

Tell us how to get an accurate bit depth reading from that (on a USB DAC) please

Binning the sample values, as mentioned. Needs a bit of trickery to deal with dithering etc but one would certainly see exactly 15 discrete bit "bands" in the histogram of the resulting stream, with a lot of empty bins in the whole 24 bit range.
Not feasible on a hardware device, as you need a lot of RAM (several GB) for this in a brute force attempt of an array of 2^24 size or even 2^32.

 

[voodooless]

Obviously a real audiophile doesn’t need the display, nor the software. The difference in bits should be like night and day