[bigshot]

Don’t understand what you’re trying to say.

 

[Audiophiliac]

There is no additional data when you upsample. It just adds more of the same. Picture your music as a number like...
1-2-3-4-5-6-7-8-9-10
and then picture it as...
1-1-2-2-3-3-4-4-5-5-6-6-7-7-8-8-9-9-10-10
It's still just 1 to 10. 11, 12 and 13 aren't in there. And if you run through it twice as fast, it's still 1 for the same amount of time. That is what upsampling does. It doesn't add data, just redundant data.

Following what you're saying, if you oversample, and your dac is busy playing what the wave actually was doing in between the samples, why doesn't it play it 2x, real fast? So that we get 200k riffs in 44.1 oversampled to 88.2?

As I've said before, the Nyquist theory (which is the foundation digital audio is built upon) says that two samples are required to PERFECTLY REPRODUCE any single frequency. You can add more points, but it won't get any better sounding. Redundancy again.

I'm still waiting until I can hear guitars strum at 100k per second, after which I can say, ok, now I want to hear each string individually.

No, because the transients that occur in music are more than 100 times slower than the time between two samples. OK, we're getting into math here, which isn't my strongest suit, but... The speed of a drum hit or decay takes up hundreds if not thousands of samples at 44.1. Remember we are talking a single sample being 1/44,000ths of a second. That is 44 times smaller than a millisecond. A fast snare drum hit from attack to peak is probably about 5 milliseconds. That is over 2,000 samples. No transient in music comes close to 1/44,000th of a second. The only thing that does is a frequency above 20kHz, which by definition would need to be have points less than 1/44,000th of a second apart? Do you follow that?

Ok, maybe it won't be that bad, but I'm still not convinced that I'll be hearing the way plastic tipped sticks strike differently than standard wood tips, without having more samples to make it out with.

Yes, CDs work with more amps and they sound just as good to human ears as higher bit/sampling rates do. CD Sound Is All You Need (see link below if you haven't clicked on it yet- all this stuff is explained better than I can explain it there.)

Someone could be saying cassette tapes copied from copies of copies is All I Will Ever Have Needed In Life, too. They reproduce everything, too.

 

[Audiophiliac]

Don’t understand what you’re trying to say.

That's what people say about mp3's. Just give them 96kbps 11khz mp3's, highest speed encoding, and tell them it sounds perfect. They won't hear anything wrong, no problem. They're like that with video, too. They like downloading 600mb copies of 2hr movies for free. That's probably why they like blockiness, too, dirty bastards. So, if I made a . on my screen, and said it was you, they wouldn't think it didn't look just like you.

 

[bigshot]

High bitrate lossy, like AAC 320 VBR is indistinguishable from lossless. I have yet to find anyone who can tell the difference. Lossy or lossless isn’t the problem to be solved, codec and data rate is.

I have a listening test I share with people. No one’s beat it yet.

 

[WoodyLuvr]

I have a listening test I share with people. No one’s beat it yet.

And they will not... unless of course they are Captain James T. Kirk!

 

[Audiophiliac]

@Audiophiliac
Here's a comparison between the same song but different sample rates. I took an audio file with a sample rate of 176400Hz. This is the actual sample rate of the recording, it's not upsampled. Keep in mind that the difference between different sample rates are the different cut-off frequency of the analog signals they can represent. So I low-passed this file with a cut-off frequency of ~20kHz. The recording still has the high resolution sample rate, I only removed the high frequencies. After that, I downsampled this file to 44.1kHz.
Here's a comparison between a slice of these file's waveform:



First of all, notice that I'm using music, not some obscure test signal. This is a very, very short slice of the music I used so you can get a better look.

There's something we have to clear up. The bottom image shows the "stairsteps" as the samples are being connected in a very simple way. This line is not what the ADC encodes. As an aside, a signal that looks like the stairstepped line would never come out of an ADC because an ADC usually puts out a PCM signal which looks nothing like that. The point of the PCM signal is not to encode information about the stairstepped signal. The output of the ADC encodes information about the continous signal which looks just like the one overlaid on top of the stairsteps.

Yeah, the stairsteps still need clearing up, that's what I expected. The adc does not encode, there is no encoding in simple pcm, it only takes a snapshot of what's playing at that point, for your dac chip to play continuously for the duration of that sample.

Something you might find more interesting is the top picture. It shows what a digital audio signal is in a less misleading way. The samples are not being connected the way they are on the bottom picture. Since a digital audio signal is discrete in time, it's a better representation. Notice how the orange line that the digital audio signal encodes is smooth. Also look at how precisely the peaks are represented. They don't have to coincide with the sample points at all, the timing of the peak does not matter, it gets encoded properly. The orange line also looks exactly like the analog voltage (if you looked with an oscilloscope, obviously noone can "see" voltage) that would come out of the DAC if you fed it the digital audio signal the white dots show.

That top wave does look nice, doesn't it. Too bad hardly anything is out at 176 yet. On Tidal it would be mqa'd to 44.1, and then flacced (which isn't nearly as bad as the mqa was, it won't remove anything or add noise to the intended result, just make a bit of noise whenever stuff happens, from the chip decoding it on the fly). Qobuz just trickles the higher res file, so you don't have to argue the mqa.

Lastly, notice how similar the top and bottom pictures are. The similarity might be unexpected to you since the music I used have a sample rate of 176400Hz and the top picture shows the 44100 signal. The reason they are this similar is because the 176400 sample is low-passed as stated before. Since it is low-passed, a digital signal with a sample rate of 44100Hz can encode it just as well as a digital signal with a sample rate of 176400Hz. The extra sample points included in the 176kHz file have no use, as the extra sample points added during the recording don't carry any extra information about the band-limited analog signal it represents.

They don't look similar. The bottom one looks like crap.

I could also send you the tested songs so you can compare them (test if they not only look the same but also sound the same) yourself by a properly conducted blind test.

Also the last couple pages of the discussion would fit better an some other threads, including this post I guess.

I've already tested getting a 176k copy of a track, after having the 44.1. Too bad they're still only starting to get more common at 96. Much improved the way analog wasn't broken, to me. Plus my dac cost a fortune compared to my analog gear, so it's analog section smokes what I had before anyways. But soundcards took a while to catch up to my first cd player. It's simple now, don't have anything go on in the computer. But, Tidal's default player sounds less bad than the other streamer's, even if it's only 44.1, and the killer deal is that it works in Audirvana, my favorite player. But so does Qobuz, that's coming next month, and streams higher res non-mqa flac.
The problem with this thread will be that only people who care about mqa will post anything. I don't, I'm just complaining that it won't stream higher res, until Qobuz comes next month. If Qobuz isn't as good with the playlist generation, I might come back and start complaining again, so that I get proper higher res for my 8 playlists, and everything else.

 

[Audiophiliac]

High bitrate lossy, like AAC 320 VBR is indistinguishable from lossless. I have yet to find anyone who can tell the difference. Lossy or lossless isn’t the problem to be solved, codec and data rate is.

I have a listening test I share with people. No one’s beat it yet.

Whenever I start thinking 'where's the beef?' about a track, I check, and sure enough, it's mp3. If I listen to above 44 for a while, going back to 44 sucks. If I listen to dsd, it's great, because I don't have much of it, so it doesn't sound normal to be that good. It took a 1 bit 2.8 million samples, can't edit because of the 1 bit, technique approach to making a file, I don't know how they manage the equivalent bit depth that way, but it works well. For tape transfers and 1-take performances, that is, so no 'each player in a studio alone, then put 'em all together' recordings, which probably sound better that way.

Aside, just a thought: If I could afford to hire my favorite electric guitarist to play here in my listening room, the portable speaker amp he brought would sound like crap compared to my system. But if he plugged straight into my amp, it would sound like I have an amazing source, and the rest would blow away what you get at live performances. I should get an even better amp...

 

[bigshot]

Following what you're saying, if you oversample, and your dac is busy playing what the wave actually was doing in between the samples, why doesn't it play it 2x, real fast? So that we get 200k riffs in 44.1 oversampled to 88.2?

Because the sampling rate is the number of samples *per second*. 44.1 or 88.2, the music plays at the same speed. 88 would be the same 44,100 samples twice... 1-1-2-2-3-3- etc.

I'm still waiting until I can hear guitars strum at 100k per second, after which I can say, ok, now I want to hear each string individually.

Ok, maybe it won't be that bad, but I'm still not convinced that I'll be hearing the way plastic tipped sticks strike differently than standard wood tips, without having more samples to make it out with.

Data rate increases won't help either of those things. Those are things that require specific miking to separate out elements in the mix.

Someone could be saying cassette tapes copied from copies of copies is All I Will Ever Have Needed In Life, too. They reproduce everything, too.

Cassettes have a significantly higher noise floor and distortion than digital audio- more limited frequency response too. Not at all the same thing.

 

[bigshot]

Whenever I start thinking 'where's the beef?' about a track, I check, and sure enough, it's mp3.

It isn't because it's an MP3, it's because it's too low a data rate MP3 and/or an old codec.

 

[VNandor]

Yeah, the stairsteps still need clearing up, that's what I expected. The adc does not encode, there is no encoding in simple pcm, it only takes a snapshot of what's playing at that point, for your dac chip to play continuously for the duration of that sample.

Where do you get your information from?The PCM signal represents the sampled analog signal. "PCM is the method of encoding typically used for uncompressed digital audio." This quote is directly taken from the wikipedia. Are you literally just making stuff up?

The stairsteps also don't need any cleaning up because there is no stairstep signal to speak of during the analog to digital conversion.

That top wave does look nice, doesn't it. Too bad hardly anything is out at 176 yet. On Tidal it would be mqa'd to 44.1, and then flacced (which isn't nearly as bad as the mqa was, it won't remove anything or add noise to the intended result, just make a bit of noise whenever stuff happens, from the chip decoding it on the fly). Qobuz just trickles the higher res file, so you don't have to argue the mqa.

The top one is at 44100Hz as evidenced by the picture and also explained in the post. Maybe take a closer look? I'm glad you're liking it though. Just show how easy it is to mistake 44100 for 176kHz

They don't look similar. The bottom one looks like crap.

The bottom one is the one with the 176kHz sample rate. It looks exactly like the top one. Try to look closer and read the post carefully this time.

On Tidal it would be mqa'd to 44.1, and then flacced (which isn't nearly as bad as the mqa was, it won't remove anything or add noise to the intended result, just make a bit of noise whenever stuff happens, from the chip decoding it on the fly).

The DAC does not have to decode FLAC on the fly since the DAC receives the PCM signal. As a result the DAC does not add noise to its output signalbecause it doesn't have to decode FLAC.

 

[Audiophiliac]

Because the sampling rate is the number of samples *per second*. 44.1 or 88.2, the music plays at the same speed. 88 would be the same 44,100 samples twice... 1-1-2-2-3-3- etc.

But, if the chip is playing what's between the samples perfectly already, what does increasing their frequency do? wouldn't it have to play each sample twice, at double speed?

Data rate increases won't help either of those things. Those are things that require specific miking to separate out elements in the mix.

Cassettes have a significantly higher noise floor and distortion than digital audio- more limited frequency response too. Not at all the same thing.

Yeah, the dark days of cassettes, where I didn't bother worrying too much about interconnects, they only sold junk in the stores. Except for that fat Monster 1'8th to rca cable I saw for my pc soundcard to stereo. My original $400 cd player sounded better than my soundcards, until my last one, which roughly equaled it, but I got a better dac and used the card's digital out anyhow. But that improved my digital, after all that time. Then, because of how well that worked out, I did it again with tons of $. Got to stop spending.

 

[Audiophiliac]

It isn't because it's an MP3, it's because it's too low a data rate MP3 and/or an old codec.

No, I only have mp3's at 320. Streaming youtube is even lower, that's way worse, but also because it's streaming.

 

[bigshot]

Where do you get your information from?The PCM signal represents the sampled analog signal. "PCM is the method of encoding typically used for uncompressed digital audio." This quote is directly taken from the wikipedia. Are you literally just making stuff up?

He doesn't know the fundamentals of digital audio and he's picking bits of info to fit his subjective guesses, but he doesn't understand enough of it to know it makes no sense whatsoever. I've been trying to speak to him on a very basic level. He understands a little bit of what I say.

 

[bigshot]

But, if the chip is playing what's between the samples perfectly already, what does increasing their frequency do?

Absolutely nothing.

wouldn't it have to play each sample twice, at double speed?

playing it twice at double speed is the same as playing it once at normal speed.

Yeah, the dark days of cassettes, where I didn't bother worrying too much about interconnects

The quality of the tape you used was pretty much the only way to improve the sound of cassettes. Also noise reduction schemes like Dolby.

 

[Audiophiliac]

Where do you get your information from?The PCM signal represents the sampled analog signal. "PCM is the method of encoding typically used for uncompressed digital audio." This quote is directly taken from the wikipedia. Are you literally just making stuff up?

Ok, I understand why that's wrong to say, pcm is a format that gets encoded to, another option could be encoding to dsd. Apologies.

The stairsteps also don't need any cleaning up because there is no stairstep signal to speak of during the analog to digital conversion.

Now you're the one making stuff up. A sample is a like a photo of audio. It doesn't include time. Snapshots are strung together so that it seems like they move, just like video has always been.

The top one is at 44100Hz as evidenced by the picture and also explained in the post. Maybe take a closer look? I'm glad you're liking it though. Just show how easy it is to mistake 44100 for 176kHz

The bottom one is the one with the 176kHz sample rate. It looks exactly like the top one. Try to look closer.

Why is the jagged one the higher bitrate one.

The DAC does not have to decode FLAC on the fly since the DAC receives the PCM signal. As a result the DAC does not add noise to its output signalbecause it doesn't have to decode FLAC.

The chip on the transport decodes the flac, that's what causes the extra noise of flac. You probably don't want to hear that there's noise from that. You may not. If you think 44.1 is perfect, and can't hear when something is an mp3, you won't hear that noise. Converting an mp3 to wav gets rid of chip decoding noise, but you can still tell there's hardly any data there. I know, mp3's still keep the perfect part, though, right?

Absolutely nothing.

playing it twice at double speed is the same as playing it once at normal speed.

According to what you're saying, playing it twice at double speed should break it, because halfway through playing it perfectly, it has to start over and play it perfectly again.

The quality of the tape you used was pretty much the only way to improve the sound of cassettes. Also noise reduction schemes like Dolby.

Maxell XLII-S. SA-90's were ok too, but the XLII-S's had a beefy heavy cassette shell, too. Dolby never worked out for me, they never changed the sound the same as other decks. Dolby turned on on my deck was not the same as in the car or on my portable, so I skipped so as to not be wrong anywhere. Plus I had dolby c on my deck, but nowhere else. It went down lower to counter more noise. Pre-recorded originals always needed it on, though. Even though what it did was not the same on all decks...