Bit Perfect Yes or No

[raindance]

Many years ago i remastered old tapes with 8bit software into CDs. Thats where bit-perfect match was most important for me, as processing would cause data loss.

On Windows, bit-perfect playback (WASAPI: Exclusive) is the easiest way how to bypass all the wrongs the Windows Audio Stack does.

I prefer bit-perfect sound only in cases when postprocessing is causing trouble, else a good EQ is better than bitperfect sound alone.

8 bit? Must have been mono recordings from am radio or a telephone answering machine. Many years ago I worked on a system that could recreate radar pulses using 4 bits, but the steps were huge and the result was an approximation at best. In that case, bit perfect wasn't perfect either ...

 

[restorer-john]

"Bit-perfect" is a marketing term, not an engineering one.

Rubbish. It absolutely is.

There is only one bit-perfect version of the Philips trio of CD test discs. The original discs. You cannot copy them, burn them, upsample, downsample them and retain the same data. The calibrated data loss cannot be copied or reproduced. The optical flatness, reflectivity and pit/land transitions are a calibration standard and cannot be duplicated.

The Pierre Verany test discs also cannot be copied as they have deliberate errors, track pitch changes and calibrated data interuptions. Impossible to reproduce other than directly from the disc itself.

I know the original industry standard pressings of a Sony YEDs or CBS CD-1 test disc are bit perfect and can be used for measuring a CD player, nothing else is verifiable. Bit perfect exists.

 

[Zensō]

Hardly. Bit perfect playback does not necessarily mean no EQ. It can be done post D/A in the analogue domain or A/D-D/A in the digital domain.

True, but the question was posed as, “This, or that?”

 

[Offler]

8 bit? Must have been mono recordings from am radio or a telephone answering machine. Many years ago I worked on a system that could recreate radar pulses using 4 bits, but the steps were huge and the result was an approximation at best. In that case, bit perfect wasn't perfect either ...

By 8bit i mean computer with Z80A CPU. Recordings were mono, with 1200, 2400 or 3600 baud and magnetic tape is ofc analogue. Using Cd quality (44KHz 16bit stereo and digital) was an overkill which guaranteed 100% success rate of loading even at highest 3600 baud rate.

Using CD audio was the most successful way of archiving old programs while keeping them as an audio track. MP3 or software players on Windows had issues,

 

[raindance]

By 8bit i mean computer with Z80A CPU. Recordings were mono, with 1200, 2400 or 3600 baud and magnetic tape is ofc analogue. Using Cd quality (44KHz 16bit stereo and digital) was an overkill which guaranteed 100% success rate of loading even at highest 3600 baud rate.

Using CD audio was the most successful way of archiving old programs while keeping them as an audio track. MP3 or software players on Windows had issues,

I misunderstood the word "programs" for some reason. Actually sounds like a very good way to back them up!

 

[luft262]

Why is it exactly that one can't use EQ, like the Wavelet App, for example, with bit perfect? Also, why is it so hard to implement bit perfect when streaming? For example, USBAPP can only implement bit perfect with Quobuz or Tidal. Why does it have to be so restrictive?

 

[Vini darko]

On android my experience is that activating bit perfect or its equivalent creates noise. I have no idea why. Its a constant hiss. The noise is low level so not always an audible thing. Unfortunately knowing it's there is enough. With UPP I use max oversample as that's silent and allows all dsp functions.

 

[ThatM1key]

In terms of serious listening I like to be "bit-perfect" as possible. For casual listening I either listen to my Magnavox CDP 610 or my Pioneer CT-W606DR. To my knowledge ripping a CD is almost bit-perfect but making a CD-R copy of a CD is not bit-perfect.

 

[Zensō]

Why is it exactly that one can't use EQ, like the Wavelet App, for example, with bit perfect? Also, why is it so hard to implement bit perfect when streaming? For example, USBAPP can only implement bit perfect with Quobuz or Tidal. Why does it have to be so restrictive?

By definition, if you change the bits in the digital domain using EQ they are no longer identical to the source, so not bit perfect.

 

[Vincent Kars]

Bit perfect is seen by many as the holy grail of computer audio.
By sending the data unaltered to the audio device, you improve sound quality!

This is of course nonsense. If nothing is altered, quality remains the same.
So where is all the fuss about?
As others in this tread already pointed out, audio quality might suffer if the OS meddles with it.
Now all OS do. They are designed with multiple audio streams in mind so all streams must be resampled (if needed) to the rate set in the audioo panel. All streams are converted to float, mixed, dithered and converted back to integer.
If you have a 16 bit DAC (or you are that silly you set the OS default to 16), you have all your sources dithered and as this is the LSB, at -96 dBFS.
If the DAC allows for 24 or even a 32 bit data path, the LSB is again dithered but who cares about -144 dBFS or -192 dBFS as no component in your playback chain is able to resolve this.

However, taking Windows , when resampling signals close to 0 dBFS, there is a measurable distortion.
This can be solved by avoiding the Win audio stack (WASAPI/Exclusive) or if you want it system wide, using EQ APO

Ending the Windows Audio Quality Debate

A debate has been long raging in the audio community that Windows isn't fit for use in a high resolution audio system. Sceptics counterargued that bits are bits and Windows audio degradations are a figment of an over-active audiophile imagination. I decided to investigate this as I kept...

www.audiosciencereview.com

So the little problems we have using Win for audio can be easily circumnavigated.

No reason to sacrifice DSP like DRC, EQ or even VC for a silly purist ideal.
The essence of EQ/DSP is to improve the sound by being as bit imperfect as hell!

 

[Jimbob54]

Why is it exactly that one can't use EQ, like the Wavelet App, for example, with bit perfect? Also, why is it so hard to implement bit perfect when streaming? For example, USBAPP can only implement bit perfect with Quobuz or Tidal. Why does it have to be so restrictive?

UAPP only works with selected services because that is what the services (or UAPP developers , or both) want. Amazon, for eg doesnt let anything other than their apps access their content. Spotify are similarly restrictive.

UAPP works by bypassing the Android audio stack and sending the signal (with DSP applied or not) directly to the DAC through its own custom driver. Other apps route the audio signal through the Android stack which automatically resamples to 48khz sample rate - meaning its no longer bit perfect then onto EQ apps like Wavelet. "Bit perfect" on UAPP requires all DSP disabled . MQA decode only works in bit perfect mode too.

This is similar in Windows- most apps go via the Windows audio stack to the DAC- Windows resamples every sound source to the same (selectable) bit depth and sample rate then EQ software like EAPO does its thing (it might be the other way round actually) then onto the DAC. Lots of music apps on PC can go bit perfect via either WASAPI exclusive mode or ASIO drivers for the DAC in question but then you cant EQ.

Roon does a bit of both and functions like UAPP on Android- you can add DSP in Roon but it will output the DSP signal in the original file format (so it upsamples to add DSP, then downsamples then outputs But its expensive

Its all a bit of a mess- but by definition you cannot have bit perfect out of the playing device in a digital format if it has any DSP added.

 

[Zensō]

Its all a bit of a mess- but by definition you cannot have bit perfect out of the playing device in a digital format if it has any DSP added.

I see what you did there.

 

[ThatM1key]

Bit perfect is seen by many as the holy grail of computer audio.
By sending the data unaltered to the audio device, you improve sound quality!

This is of course nonsense. If nothing is altered, quality remains the same.
So where is all the fuss about?
As others in this tread already pointed out, audio quality might suffer if the OS meddles with it.
Now all OS do. They are designed with multiple audio streams in mind so all streams must be resampled (if needed) to the rate set in the audioo panel. All streams are converted to float, mixed, dithered and converted back to integer.
If you have a 16 bit DAC (or you are that silly you set the OS default to 16), you have all your sources dithered and as this is the LSB, at -96 dBFS.
If the DAC allows for 24 or even a 32 bit data path, the LSB is again dithered but who cares about -144 dBFS or -192 dBFS as no component in your playback chain is able to resolve this.

However, taking Windows , when resampling signals close to 0 dBFS, there is a measurable distortion.
This can be solved by avoiding the Win audio stack (WASAPI/Exclusive) or if you want it system wide, using EQ APO

Ending the Windows Audio Quality Debate

A debate has been long raging in the audio community that Windows isn't fit for use in a high resolution audio system. Sceptics counterargued that bits are bits and Windows audio degradations are a figment of an over-active audiophile imagination. I decided to investigate this as I kept...

www.audiosciencereview.com

So the little problems we have using Win for audio can be easily circumnavigated.

No reason to sacrifice DSP like DRC, EQ or even VC for a silly purist ideal.
The essence of EQ/DSP is to improve the sound by being as bit imperfect as hell!

The amount of times I reinstall windows, I don't get proper ASIO support with Topping's ****** drivers. With PCM content it leaks ASIO audio in the WS audio unless you disable it in the sound panel manually. If you don't install the topping drivers and use the "plug n play" drivers, you don't get 44.1khz 16bit support, only 44.1khz 24bit. The "plug n play" drivers will reject 16bit 44.1khz WASAPI Exclusive requests.


I have a thread about this here:

Topping ASIO Leak Problem

I searched hard on google and couldn't find a single answer. I always get results about people asking why ASIO blocks another programs sounds but ironically my problem is the opposite. The program I use is foobar2000 So what happens is when I force ASIO on my Topping E30, the "ASIO" says it...

www.audiosciencereview.com

 

[ZolaIII]

Why is it exactly that one can't use EQ, like the Wavelet App, for example, with bit perfect? Also, why is it so hard to implement bit perfect when streaming? For example, USBAPP can only implement bit perfect with Quobuz or Tidal. Why does it have to be so restrictive?

Wavelet is a DSP-EQ which ties to default Android OS audio chain (at the end of it before output) and there for it can't tide to costume in app chain which uses it's own output (USB Audio 2.0 app driver) disregarding of bit perfect. Bit perfect disable any processing anyway.
Usually propetry (IP) bit streams end up being a "bit perfect" but not because that means anything meaningful but because you can't add anything (closed source decoding chain) exampl; Dolby's, DSD and MQA.

 

[Kegemusha]

In my PC, In foobar, you can have bit perfect, but, I could be wrong, I think I can still do some EQ in foobar.

 

[Freeway]

I prefer bit perfect to begin with.

 

[sonci99]

Bit perfect is beautiful..

I remember the test used at the time. You play a dts file to the spidf out of a soundcard, to a receiver, and if the dts logo lights than your playback is bit perfect..

 

[Berwhale]

Rubbish. It absolutely is.

There is only one bit-perfect version of the Philips trio of CD test discs. The original discs. You cannot copy them, burn them, upsample, downsample them and retain the same data. The calibrated data loss cannot be copied or reproduced. The optical flatness, reflectivity and pit/land transitions are a calibration standard and cannot be duplicated.

The Pierre Verany test discs also cannot be copied as they have deliberate errors, track pitch changes and calibrated data interuptions. Impossible to reproduce other than directly from the disc itself.

I know the original industry standard pressings of a Sony YEDs or CBS CD-1 test disc are bit perfect and can be used for measuring a CD player, nothing else is verifiable. Bit perfect exists.

The OPs question has nothing to do with test CDs or CD pressed to make copying difficult. The OPs question relates to audio data, the things you mention above relate to the physical media, not the data encoded upon it.

 

[ZolaIII]

The OPs question has nothing to do with test CDs or CD pressed to make copying difficult. The OPs question relates to audio data, the things you mention above relate to the physical media, not the data encoded upon it.

Well not quite as such, as owners of such IP's are the ones who propagate the most "bit perfect" as a gift from heaven and if majority of people figured out how that's a snake oil (claiming how preventing you to access something in order to reprocess and correct it to your needs makes it sound better as a bit perfect) they either couldn't sell any of those or had to open them up to the extent that you actually could insert your plug in (DSP - EQ) into the chain.

Disclaimer I do want ideally 21 bit integer precision PCM bit perfectly written (and recorded initially as such) and written only (as for storage disregarding of packaging/data compression).

 

[radix]

I don't understand the "bit perfect" world, especially the "I want what came out of the mic" sentiment. I did some work in radio back in the 80s-90s, but I'm not in the industry. This is my understanding, and if I'm wrong please correct me.

• Recordings are engineered and EQ'd and level mixed. You won't get what came out of the mic. Modern pop is highly engineered, sometimes with 100s of takes spliced together for pitch accuracy (instead of using a machine). There's a whole processing chain for each track, like vocal HP filters, compression, reverb, etc.
• Your headphones and speakers/room are not transparent. You need to EQ them such that what you hear is close to what is on the recording.
• If you EQ in the digital domain, I believe you need to translate 1-bit into multi-bit (i.e. PCM), and your EQ is going to swamp whatever refinement (if any) was in the DSD vs PCM.
• If you EQ in the analog domain... well, bit-perfect is lost.
• Whenever you quantize a sample to a value, you have error. This is minimized by dithering, so again the bits you get are not what were sourced. I'm not sure what DSD does for dithering?

On a broader level, music, like photography, is about art not about technically precise reproduction of the source (with a few exceptions). Mics have color, the studio has color, the mix has color, the production has color, your gear has color. The goal is to produce a product the evokes the desired emotion or feeling in the listener/viewer. Who wants to see a photo straight off the sensor?