1. Do you experience a big difference when running PianoTeq in 44100 versus 48000 hz mode? I feel like high notes played on the K1 sound waaay better - more defined, less cottony - in 48000 hz mode.
2. Which is your favorite midi sequencer to run PianoTeq from? As a sidenote; does it support 48000 hz?
Thanks,
Anders.
1. Do you experience a big difference when running PianoTeq in 44100 versus 48000 hz mode? I feel like high notes played on the K1 sound waaay better - more defined, less cottony - in 48000 hz mode.
2. Which is your favorite midi sequencer to run PianoTeq from? As a sidenote; does it support 48000 hz?
Thanks,
Anders.
Yeah, there is a big difference between 41000 and 48000!!! Very noticeable Anders.
There is much more 'character' in 48000 than 41000.
Naturally, because Nyquist limit is shifted upwards from 22050 to 24000 Hz (good luck in hearing those ). However, higher sample rate probably allows Pianoteq to render higher partials with much more precision, hence you perceive it as sounding "better".
Golly, even I can hear it (I think). Thanks for the tip!
Interesting!
Did any of you ever do a blind test?
Um, how do I run Pianoteq in 48000 mode? 44100 seems to be the highest option on my copy (3.6.6)...
Well, your soundcard/ASIO driver need to support higher sample rates...
Well, your soundcard/ASIO driver need to support higher sample rates...
That would explain it. And there I was feeling all smug that because I had a mac I didn't need to bother with all this soundcard malarky I read so much about on this forum. Looking online for a nice USB audio interface for my Macbook pro. Any recommendations anyone? And while I'm at it, should I be exploring 5.1 output? Has anyone tried using the 5 mic feature on Pianoteq to create surround sound? btw, Anders, I assume that Pianoteq doesn't support 96000 yet, although it might one day?
It does support 96k, as far as I know.
There are some users (jfelice88keys, for example) who used the microphone positioning to create surround sound.
I assume that Pianoteq doesn't support 96000 yet, although it might one day?
It does, I just tried it. You have to change the sample rate in the "Devices" section of Pianoteq.
JR
Just to avoid any possible confusion: Pianoteq internal sample rate can be set to 96 kHz only in the pro version. Of course a 96 kHz external sample rate can be used with all Pianoteq versions, but in standard or play versions, the internal sample rate (used by the model for the computations, adjustable in the Options/"Perf" section) will be limited to 48 kHz. Taking into account our ears limitation and the Niquist frequency mentioned by EvilDragon, that should be more than sufficient, although I am aware of the very intense debates that can take place around this question.
Yeah, there is a big difference between 41000 and 48000!!! Very noticeable Anders.
There is much more 'character' in 48000 than 41000.
Hello All,
[EDIT:] I am not here to agree or dispute whether there is a big difference in audio heard, sampled at 48kHz or 44.1kHz. Rather, I am here to present to you a fascinating Youtube video about the human perception system. The video is a 56-minute condensation of a six-hour forum of the Audio Engineering Society in 2009, entitled 'Debunking Audio Myths.' [end EDIT]
http://www.youtube.com/watch?v=BYTlN6wjcvQ
This Youtube video has been mentioned before, but the point is ... we can all be subconsciously "steered" to hear ... or not hear ... differences -- EVEN WHEN PRESENTED WITH THE SAME STIMULUS! The problem is: If you "expect" to hear a difference, you WILL hear a difference. Conversely, if you are convinced there is no difference, you will steer yourself to hear no difference.
One example is an anecdote of how the presenter wired a plain wooden box with a toggle switch on it. One side was labeled "Tube", and the other side was labeled "Transistor". Present in the room was a Macintosh amplifier (a "dead" amp salvaged from a junk heap, which had fake tubes lit inside of it), and a cheap 30-watt electronic amplifier. With thirty people each trying this box -- ostensibly to switch between the Macintosh and the cheap electronic amplifier -- the "audiophiles" unanimously preferred the tube sound, claiming that the Tube setting sounded warmer; mostly the electrical engineers preferred the transistor sound, claiming they liked the clarity, for their own reasons.
In the end, the narrator divulged that the toggle switch was not connected to anything!
* * * * *
In another presentation, a young lady presents a tape of some soft-rock music played backwards. It is not clearly discernible what the lyrics were "saying" in the first go-'round of the tape. But, when she projected some words for everyone to read along, then the SAME audio was amazingly able to be understood!
* * * * *
While this is nearly an hour-long video, it is well worth your saving it as an entertaining reference for your own pleasure, and for others. By the way, audio examples of various bit-rates, dither types, and purposefully introduced distortion, are included in the video.
Cheers,
Joe
Thanks Joe - very interesting.
In first year engineering during a physics lecture, the prof brought out a sound generator. He had all of us stand up, and started the generated frequency quite low (perhaps around 512 Hz), and gradually increased the frequency. When we could no longer hear the sound, he asked that we sit down.
The lecture theater held about 175 students, all about 18 years old, so very few of us had suffered any job or age related hearing loss.
By the time the frequency was at 18,000 Hz, only a few of us were left standing. At 20,000, there was one student in the whole theater left standing. Note that we were all males (females have better high frequency hearing, but just didn't enter engineering in the late fifties). I dropped out below 18,000 Hz.
The professor then pointed out that by the time males had reached adolescence, they had little or no hearing beyond 18,000 Hz. Of course his test had already demonstrated that.
Has anyone used a wave editor and looked at the frequencies generated by a piano on good CD recording of a piano? Tell me what the highest frequency was, and whether or not the level of these frequencies was significant?
Glenn
Interesting posts, Joe and Glenn!
That's why I asked about a blind test...
I know this is kinda controversial, but personally, I don't believe that anyone will hear a 'big' difference between 44.1 and 48 kHz sample rates.
Anyway, the following suggestion by EvilDragon might be an explanation:
However, higher sample rate probably allows Pianoteq to render higher partials with much more precision, hence you perceive it as sounding "better".
Philippe, could you maybe briefly comment about that?
Does the (internal) sample rate of Pianoteq in any way affect the computation "within" the model?
Thanks!
Wolfgang
I know this is kinda controversial, but personally, I don't believe that anyone will hear a 'big' difference between 44.1 and 48 kHz sample rates.
This is the way I look at it...
Imagine the difference between images at low resolution compared with images at a higher resolution...the higher resolution looks better (i.e. smoother, less jaggies, etc.) and it is the same (more or less) with audio samples. The jagged bits are still there, but they are doubled, quadrupled, etc. to 'smooth' things out.
Higher resolution is more information that equates to better quality...generally. For instance, you can't take a low resolution sample and just double the resolution thereby doubling the quality. You are just doubling the information. You need to capture that information at a higher resolution in the first place.
Furthermore, rendering a plugin at a very high resolution may not get you the results you are looking for. A plugin that does analog emulation of a Moog synth (for ex.) may not sound as good rendered at 96khz...the 'lower resolution' render may actually sound more authentic. It also depends on how the plugin was coded too (the algorithms).
You have to do A/B comparison tests. Render at 44khz and then at 48khz, 96khz, etc. and use waveform tools and your ears to determine if things are indeed better.
Maybe someone could explain it better...
JR
Philippe, could you maybe briefly comment about that?
Does the (internal) sample rate of Pianoteq in any way affect the computation "within" the model?
Thanks!Wolfgang
The internal sample rate of Pianoteq affects the computation within the model in the following way: if F denotes the internal sample rate, that means that all internal audio computations made by the model are done at the rate F and there cannot be any information above F/2 (the Nyquist frequency).
You can hear this very well on the harpsichords: this instrument is richer in overtones than the piano, it has overtones that go easily up to 20 kHz. No problem if you use 44.1 kHz because our ears are anyway limited to 20 kHz (and you need to be very young to hear that high!). But if you turn the internal sample rate to let’s say 22kHz, then the model won’t compute anything above 11 kHz, and you will be missing all spectral components lying from 11 kHz to 20 kHz: that makes really a clearly audible difference. Regarding the piano, it hardly contains significant partials above 15 kHz, which explains why you will have it difficult to hear a difference between 29.4 kHz and 44.1 kHz internal sample rate.
Thanks for clearing things up, Philippe!
Regarding the piano, it hardly contains significant partials above 15 kHz, which explains why you will have it difficult to hear a difference between 29.4 kHz and 44.1 kHz internal sample rate.
This is what I was questioning - with my wave/audio editor, I rarely if ever see any significant frequencies above 16 kHz in a spectral analysis.
Let me put this forward for discussion - the soundboard of a piano is about 3/8" inch (9 mm) thick - I really suspect that the soundboard of a violin is much thinner - and the length of the violin's soundboard is much shorter than that of a piano. From basic physics, it would be reasonable for a violin to have much higher frequency partials than a piano.
I suspect that there is a higher probability that the strings have more high frequency components than the wood soundboard, but if anyone thinks the strings have a major effect on piano sound, then compare the sound levels of an acoustic guitar to a solid body guitar with the amplifier turned off. The sound of a good acoustic guitar can fill an auditorium with several hundred people - a solid body guitar with no amp can't fill a small bedroom.
I wouldn't suggest that very high frequencies don't exist in a piano's sound, but I would suggest that:
1) they aren't significant (please forget what audio sales people say),
2) very few, if any of us (particularly males) can hear them (I know some like to think they can ).
3) the placebo effect is much stronger than most of us would like to admit (I listened to the entire U-tube link that Joe posted, and honestly I wasn't surprised by anything that was said).
Glenn
The soundboard works like a linear amplifier (a time invariant linear system in signal theory is called a filter), thus the frequencies found in the output are frequencies that were already present in the input and were amplified by a certain factor (called the transfer function in signal theory – sorry, I don’t want to look pedant here but just want to show how universal the question is).
The amplification factor will depend on the resonance frequencies of the soundboard: high when close to such a frequency, lower when far. But whatever the size of the soundboard, it has resonance frequencies up to infinity, and moreover, these resonance frequencies have in mean value a uniform density from 0 Hz to infinity. The difference between a small and a big soundboard will be that the big soundboard will have its first resonance frequency lower than a small soundboard (in case of equal thickness) – and subsequently show a higher resonance frequencies density.
That means in particular that small and big soundboards are equally capable of amplifying high frequencies, but big soundboards are more adequate to amplify low frequencies.
So the reason why there are high frequencies is not to find in the soundboard itself but in the way it has been excited. And that’s why a harpsichord is richer in high overtones: a string plucked by a plectrum produces much stronger overtones than the same string stroked by a hammer covered with felt.
I wasn't aware that the soundboard was an amplifier - I thought it was a transducer. Better get my old physics books out.
To Glenn and all who are interested,
Glenn, when you mentioned reading your old physics books, it reminded me of a great book that I have in my library called Science and Music by Sir James Jeans. It is dated so far as technology is concerned (first publication in 1937) but is in my opinion a classic text. It marries the love of music, artistic performance, and subjective hearing of music, with the scientific method and accurate physical analysis. Everything ranging from soundwaves, harmonics, soundboards, music theory, music history, human hearing, to hall acoustics is discussed.
It won't be nearly as technical as your physics texts are, but it has the perfect blend of our hearing of music and the science behind it.
I highly recommend it. :-)
To Glenn and all who are interested,
Glenn, when you mentioned reading your old physics books, it reminded me of a great book that I have in my library called Science and Music by Sir James Jeans. It is dated so far as technology is concerned (first publication in 1937) but is in my opinion a classic text. It marries the love of music, artistic performance, and subjective hearing of music, with the scientific method and accurate physical analysis. Everything ranging from soundwaves, harmonics, soundboards, music theory, music history, human hearing, to hall acoustics is discussed.
It won't be nearly as technical as your physics texts are, but it has the perfect blend of our hearing of music and the science behind it.
I highly recommend it. :-)
1937 you say - might be a hard one to find. I thought I was older than dirt, but this book is actually a few years older than I am (man, I took my first piano lessons in 1946 - time flies when you're having fun).
Sounds interesting - might be in our local up to date library. Most of the books on photography are about film.
Glenn
I wasn't aware that the soundboard was an amplifier - I thought it was a transducer. Better get my old physics books out.
You are perfectly right of course, it is a transducer. The word amplifier was more a metaphore illustrating what you said: the fact that the sound is louder with a soundboard than without. The main point is that the frequencies that you can find in the output of the transducer are due to the input, and are more or less amplified by the transducer.
wolfgang wrote:Philippe, could you maybe briefly comment about that?
Does the (internal) sample rate of Pianoteq in any way affect the computation "within" the model?
Thanks!Wolfgang
The internal sample rate of Pianoteq affects the computation within the model in the following way: if F denotes the internal sample rate, that means that all internal audio computations made by the model are done at the rate F and there cannot be any information above F/2 (the Nyquist frequency).
It's true, but that's not the only way the computation could be affected by the internal rate. Even if you filter down to 44 kHz at the end, it can still improve sound to oversample internally, for example because of aliasing. Are Pianoteq's algorithms immune to this?
To the original poster: if you really can hear a difference then you should record the same MIDI file with both settings and see if you can still hear the difference.
To Glenn and all who are interested,
Glenn, when you mentioned reading your old physics books, it reminded me of a great book that I have in my library called Science and Music by Sir James Jeans. It is dated so far as technology is concerned (first publication in 1937) but is in my opinion a classic text. It marries the love of music, artistic performance, and subjective hearing of music, with the scientific method and accurate physical analysis. Everything ranging from soundwaves, harmonics, soundboards, music theory, music history, human hearing, to hall acoustics is discussed.
It won't be nearly as technical as your physics texts are, but it has the perfect blend of our hearing of music and the science behind it.
I highly recommend it. :-)
Thanks for mentioning this book. It looks interesting, and I was able to find it on Amazon in the U.S.
Hi Michael,
That's great that it is still available from Amazon. I think I bought my edition a while back at Borders bookstore. Like I mentioned, some of the examples are a little out-of-date (there are mentions of gramophone recordings) but since it uses scientific data on the physics of music making and acoustics, most of the book is relevant today and even timeless.
Philippe Guillaume wrote:The internal sample rate of Pianoteq affects the computation within the model in the following way: if F denotes the internal sample rate, that means that all internal audio computations made by the model are done at the rate F and there cannot be any information above F/2 (the Nyquist frequency).
It's true, but that's not the only way the computation could be affected by the internal rate. Even if you filter down to 44 kHz at the end, it can still improve sound to oversample internally, for example because of aliasing. Are Pianoteq's algorithms immune to this?
I believe yes, we do our best efforts to avoid any aliasing throughout the whole process.