Solid Snake-Oil Storage: This SSD Is Aimed at Audiophiles

[raghupb]

Have a very basic question. Someone with a good understanding of PC based audio can pitch in.
For audio, specifically 16/44.1 compressed FLAC or uncompressed WAV, do we need a high speed drive?
I can understand SSD is useful for laptops/desktops to improve on memory read/write speed in a high performing compute environment.

Cheers,
Raghu

 

[firearm12]

1. The Allo USBbridge has its own DAC and the output of that is what you hear.

Usbridge does not have any dac inside. Usbridge card is connected to rpi and just offers a different electrically clean usb port which connects to an external dac using usb cable. Functionally Its just a normal usb port similar to any other usb port on rpi. It's supposed to transfer same zeroes and ones as they will be transferred if the dac is connected to any other usb port on rpi. Infact you can use the usbridge usb port as any other usb port, its not made specially for audio data transfer only. My point is that the way zeroes and ones are transferred, it does matter. It can matter in case of hard disks also.

 

[raghupb]

1. You are comparing two completely different hardware. In the Allo USBbridge, the RPi is used as just a computing and processing device.
2. It still uses the same hard disk, SD card, and RAM as you would use with the RPi. So there is no difference there. Our discussion here is whether a 'better' hard disk would make a difference. Is the Allo USBBridge offering you or telling you to use an improved hard disk? They are not. They dont care what hard disk you use. They also go with whatever SD card and RAM that the RPi uses.
3. The Allo USBbridge has its own DAC and the output of that is what you hear.
4. Have you ever taken the output from a stock RPi through the USB, fed it to a properly configured external DAC and listened to the sound?
5. The RPi's internal DAC is an inexpensive low quality DAC, and that makes a huge difference. The requirement of the RPI design did not justify an expensive DAC.
6. And remember, at the end of the day, the Allo USBBridge uses the stock data path and connectors that comes with the RPi. What they do is to ensure that, beyond that point, the data is processed properly, DACed well and sent to your ears. The PRI does send the data block from the HDD, SSD, SD or RAM perfectly to the USBBridge. If it does not, you will still hear noise how much ever the Allo dances.
7. As I have said many times before, there can be no doubt on the digital storage and processing. It is what happens to the data after it leaves the digital domain that matter.
8. The same RPi processes and displays pristine 4K video with mind boggling clarity and full 7.1 digital surround sound. You think it is going to be bothered by a piddling two channel sound?

Cheers

Allo USBridge and Digione are just transports. The former is used to connect to USB input on DAC and the latter to a COAX S/PDIF input.
I'm currently, running my Digione RPI USB out to USB in of a Soekris DAC.
I don't hear any "gremlins" or I'm stone deaf (which I partly am )

Cheers,
Raghu

 

[sidvee]

Have a very basic question. Someone with a good understanding of PC based audio can pitch in.
For audio, specifically 16/44.1 compressed FLAC or uncompressed WAV, do we need a high speed drive?
I can understand SSD is useful for laptops/desktops to improve on memory read/write speed in a high performing compute environment.

Cheers,
Raghu

I have compared FLAC's ,WAV and Hi rez/DSD files stored on both SSD and HDD. I could not hear any difference what-so-ever. Yes write speeds make it a less time consuming affair to transfer files to the SSD (and whatever all the other benefits are), but from a SQ perspective I couldn't really hear anything.
If there is a difference, then perhaps my system is not revealing enough. Maybe the Nvme ssd in the OP will show the difference, who knows.
Cheers,
Sid

 

[keith_correa]

For audio, specifically 16/44.1 compressed FLAC or uncompressed WAV, do we need a high speed drive?

No "need". Most players worth their salt, buffer the track into memory and then read from data buffered into memory. But if memory limited, then yes, sequential I/O speed would/might matter. But even if that is the case, there are ways that can be mitigated by read ahead methods. But that is a black box with commercial players.

 

[raghupb]

I have compared FLAC's ,WAV and Hi rez/DSD files stored on both SSD and HDD. I could not hear any difference what-so-ever. Yes write speeds make it a less time consuming affair to transfer files to the SSD (and whatever all the other benefits are), but from a SQ perspective I couldn't really hear anything.
If there is a difference, then perhaps my system is not revealing enough. Maybe the Nvme ssd in the OP will show the difference, who knows.
Cheers,
Sid

Agree with this, but such transfers are few and far between (at least in my case)

No "need". Most players worth their salt, buffer the track into memory and then read from data buffered into memory. But if memory limited, then yes, sequential I/O speed would/might matter. But even if that is the case, there are ways that can be mitigated by read ahead methods. But that is a black box with commercial players.

The Digione has 2 parameters
"Audio Buffer Size" which is set to 2MB by default
"Buffer Before Play" which is set to 10% by default
I guess it should be good enough for normal listening as I've never changed them.

Cheers,
Raghu

 

[superczar]

Have a very basic question. Someone with a good understanding of PC based audio can pitch in.
For audio, specifically 16/44.1 compressed FLAC or uncompressed WAV, do we need a high speed drive?
I can understand SSD is useful for laptops/desktops to improve on memory read/write speed in a high performing compute environment.

Cheers,
Raghu

+1 to what Keith Said.
No matter how high res the file.


To put it in perspective, a completely uncompressed 24 bit 192 khz sampling rate file will be about 70MB for a minute.
That translates to a little over 1 MB read speed a second.

Current nvme drives do 6-8GB/s (6000x-8000x of what is needed) sequential read speeds
Older SSDs do maybe 1GB/s
Even the oldest spin drives do at least 100 MB/s (100X)

Audio playback is very very far from requiring a high performance compute environment
for the lack of a better analogy, vanilla audio playback (even high res) will tax a modern PC the same way as a race against a bicycle for a Veyron

 

[venkatcr]

Usbridge does not have any dac inside. Usbridge card is connected to rpi and just offers a different electrically clean usb port which connects to an external dac using usb cable. Functionally Its just a normal usb port similar to any other usb port on rpi. It's supposed to transfer same zeroes and ones as they will be transferred if the dac is connected to any other usb port on rpi. Infact you can use the usbridge usb port as any other usb port, its not made specially for audio data transfer only. My point is that the way zeroes and ones are transferred, it does matter. It can matter in case of hard disks also.

I stand corrected on the DAC. I seemed to watched the wrong video on ASR. It was BOSS2 and no USBBridge. My bad.

At the same time I simply give up on the statement ' My point is that the way zeroes and ones are transferred, it does matter. It can matter in case of hard disks also.'

I am so glad reality is different. Other wise planes will be falling from the sky, all our hard earned money would simply disappear, equipment in hospitals will pump the wrong medicines into patients or give the wrong diagnosis, and the IT department will continue to harass you with wrong information.

As someone said, please do believe whatever you want. It really does not matter.

Cheers

 

[arj]

Usbridge does not have any dac inside. Usbridge card is connected to rpi and just offers a different electrically clean usb port which connects to an external dac using usb cable. Functionally Its just a normal usb port similar to any other usb port on rpi. It's supposed to transfer same zeroes and ones as they will be transferred if the dac is connected to any other usb port on rpi. Infact you can use the usbridge usb port as any other usb port, its not made specially for audio data transfer only. My point is that the way zeroes and ones are transferred, it does matter. It can matter in case of hard disks also.

I presume you mean this where there is a timing issue around it eg associated with a clock and consumption is based in that as well ie when the sequence and timing are important. if that is not an issue then it should not matter eg my understanding of protocols like TCP/IP etc are all packets of data which come together and only then out together

 

[raghupb]

@arj
Where is TCP/IP in a simple HDD (FLAC/WAV) --> Player (on processor) --> Audio out (PCM)?

In streaming, I agree. But media streaming uses UDP not TCP as re-transmission after a set delta does not make sense.
Even if TCP/IP is used, it is working at 100Mbps and operates at block (or window) level.
Audio which needs a few Mbps, as @superczar mentioned will still not be affected.

If I've got this correct, HDD/storage read/retrieval is bit perfect.
If there is noise due to a poor PS or very high speed switching ASICs/memories, at most an LSB or two may get buried under.
So the DAC (if operating at 16 bit resolution) effectively becomes 15- or 14- bit.
This is so minuscule for human ears to discern, isn't it?

Cheers,
Raghu

 

[venkatcr]

Arj, this is just adding to the confusion. USB, Serial, Parallel, Optical or any other mode are very simple transport mechanisms that just move packets of data from one point to another. Some are two way while some are one way. The two way mechanism is used where the data packet's authenticity has to be verified, and also where there is some sort of control or other instructions are to be returned to the sender.

Modern DACs, that use either USB or Optical, use buffers to store the data packets locally and employ their own clocks when converting the data. The transport (and consequently the storage) mechanism(s) have very little chance of affecting the results.

In our mad rush to get better audio, we are all just chasing ghosts, trying to find faults where none exist.

 

[firearm12]

Modern DACs, that use either USB or Optical, use buffers to store the data packets locally and employ their own clocks when converting the data. The transport (and consequently the storage) mechanism(s) have very little chance of affecting the results.

This is wrong imo. Transport mechanisms has most impact on sound quality. Buffers don't help resolve the problem as music is played in real time as soon as you hit play button and they work on mechanisms where there is a rate at which data goes in and a rate at which it comes out of buffers. If there is even a small difference in these rates, buffers will overflow or get empty. Output needs to track the input in order to play audio without any sync and delay issues, this tracking don't allow buffers to eliminate all problems. On top of all this is problem of clean current and imperfect clocks.

 

[arj]

@arj
Where is TCP/IP in a simple HDD (FLAC/WAV) --> Player (on processor) --> Audio out (PCM)?

In streaming, I agree. But media streaming uses UDP not TCP as re-transmission after a set delta does not make sense.
Even if TCP/IP is used, it is working at 100Mbps and operates at block (or window) level.
Audio which needs a few Mbps, as @superczar mentioned will still not be affected.

If I've got this correct, HDD/storage read/retrieval is bit perfect.
If there is noise due to a poor PS or very high speed switching ASICs/memories, at most an LSB or two may get buried under.
So the DAC (if operating at 16 bit resolution) effectively becomes 15- or 14- bit.
This is so minuscule for human ears to discern, isn't it?

Cheers,
Raghu

I meant HDD streaming vie Ethernet/wireless to a streamer..since as Venkatcr mentioned these are buffered and the packet regroup.

 

[mbhangui]

This is wrong imo. Transport mechanisms has most impact on sound quality. Buffers don't help resolve the problem as music is played in real time as soon as you hit play button and they work on mechanisms where there is a rate at which data goes in and a rate at which it comes out of buffers. If there is even a small difference in these rates, buffers will overflow or get empty. Output needs to track the input in order to play audio without any sync and delay issues, this tracking don't allow buffers to eliminate all problems. On top of all this is problem of clean current and imperfect clocks.

@firearm12 has got it. My understanding almost matches with the above. Playing music is not same as copying a file from one directory to another or from one machine to another. It also inolves time domain which people keep forgetting. Copying hard disk is bit perfect. It doesn't matter if the file copy happened in 1 second or 1.1 seconds. After the copy you will get the same file as the original.

Music playback is very different. Let us for the moment forget about the sampling rate and assume that after delta sigma conversion we got a perfect representation of the original signal. So let us just consider that we are going to play music which has been perfectly copied to a wav, flac, dsd file. Now for the point of simplicity let us assume that the music is a 50Hz tone. This means for the music to be perfect, a sine wave has to alternate between plus and minus exectly 50 times in exactly 1 second. You Play the 50 Hz in 1.1 second, you just lowered the frequency. You play the 50 Hz in 0.9 second, you have increased the frequency. Because of the noise and also an imperfect clock, any dac will nevery be able to play 50 Hz exactly continuously. To achieve perfect playback, an insane amount of engineering goes into the dac to engineer at minimum the below

1. Very high quality clocks
2. Very stable oscillators based on the above clocks. These oscillators are affected by noise and power supply
3. Extremely stable power supplies using large caps, super caps and low drop out voltage regulators
4. Insane amount of shielding
5. R&D for delta signma algorithms.

Things which where so easy in a LP player (because of heavy platter making it difficult to alter the RPM suddenly, stable mains frequency, etc) now involves better engineering, better components to excel the sound that came from a decent LP player.

 

[raghupb]

Boy-o-boy!!
This business sounds like rocket science. I've been listening to computer based music for over 25 years.
Never had I imagined it to be so complicated.

Anyways, what the DAC does has nothing to do with HDD + player.
This is what the original discussion is all about, fancy SSD vs normal SSD vs spinner HDD.

Cheers,
Raghu

 

[keith_correa]

Among all the outrage, the eagerness to explain why it cannot work, the eagerness to explain why the "philosophy" behind it is sound (pun intended) et al., did we miss the fact that this is supposed to and meant to host the OS and not music files?

Someone rightly named the likes of us: "Audiophools". <insert evil sounding laugh here>

 

[reignofchaos]

I have compared FLAC's ,WAV and Hi rez/DSD files stored on both SSD and HDD. I could not hear any difference what-so-ever. Yes write speeds make it a less time consuming affair to transfer files to the SSD (and whatever all the other benefits are), but from a SQ perspective I couldn't really hear anything.
If there is a difference, then perhaps my system is not revealing enough. Maybe the Nvme ssd in the OP will show the difference, who knows.
Cheers,
Sid

You are not alone Sid. Tried hdd, sata SSD, nvme SSD, usb drives. Nothing made any difference. Maybe I don't have golden ears or my equipment is too poor.

 

[superczar]

@firearm12 has got it. My understanding almost matches with the above. Playing music is not same as copying a file from one directory to another or from one machine to another. It also inolves time domain which people keep forgetting. Copying hard disk is bit perfect. It doesn't matter if the file copy happened in 1 second or 1.1 seconds. After the copy you will get the same file as the original.

Music playback is very different. Let us for the moment forget about the sampling rate and assume that after delta sigma conversion we got a perfect representation of the original signal. So let us just consider that we are going to play music which has been perfectly copied to a wav, flac, dsd file. Now for the point of simplicity let us assume that the music is a 50Hz tone. This means for the music to be perfect, a sine wave has to alternate between plus and minus exectly 50 times in exactly 1 second. You Play the 50 Hz in 1.1 second, you just lowered the frequency. You play the 50 Hz in 0.9 second, you have increased the frequency. Because of the noise and also an imperfect clock, any dac will nevery be able to play 50 Hz exactly continuously. To achieve perfect playback, an insane amount of engineering goes into the dac to engineer at minimum the below

1. Very high quality clocks
2. Very stable oscillators based on the above clocks. These oscillators are affected by noise and power supply
3. Extremely stable power supplies using large caps, super caps and low drop out voltage regulators
4. Insane amount of shielding
5. R&D for delta signma algorithms.

Things which where so easy in a LP player (because of heavy platter making it difficult to alter the RPM suddenly, stable mains frequency, etc) now involves better engineering, better components to excel the sound that came from a decent LP player.

And where exactly does the HDD come into this?

As for your argument on DACs and time domain drifts, Take a cheap DAC like the Apple USB C dongle (sub 1K INR?)
use any waveform generator to create a 50hz (or any other frequency) test tone
Feed analog output into a half decent oscilloscope - and check if you see any time domain problems
(You won't)

Believe it or not, audio DACs sit at a rather low end of the engineering complexity spectrum of digital signaling

 

[mbhangui]

And where exactly does the HDD come into this?

Considering the HDD is connected on the same system to which the DAC is connected and the noise generated affects the USB on which the DAC is connected.

 

[superczar]

Considering the HDD is connected on the same system to which the DAC is connected and the noise generated affects the USB on which the DAC is connected.

That has nothing to do with the time axis though now, does it?