But does anyone know of a site that has a set of test recordings done in all of the formats, for comparison? There must be several.
]]>FWIW, here's someone's MP3 test results:
http://www.mp3-tech.org/tests/gb/index.html
(not sure of their credibility)
Here's a web page which talks about double blind testing, with references to software that allows one to do their own double blind testing:
http://en.wikipedia.org/wiki/ABX_test
I have not tried the software.
Greg.
]]>Same thing with pianos, but I don't notice it as much unless I hear a wave file and an mp3 file of the same piece side by side. The volume level seems to go down too, but I'm not sure why.
]]>Greg.
]]>so does that mean that an MP3 is a reasonable format for listening to solo piano work (realizing that recording techniques vary greatly...)?
MP3s use standard sample rates of 44.1kHz or 48kHz, so you don't necessarily lose any high frequencies. Depending on the bitrate though, the frequencies you do hear will be variably quantized or discarded, giving rise to the reduced audible quality. The higher the bitrate, the better it sounds. Also, the simpler the signal, ie the less "busy" the spectrum, the easier it is to encode more accurately, and the better it will sound. So some pure and simple solo piano music might encode very well at moderate bitrate. The quality of the encoder matters too!
]]>I'm thinking about upgrading to Pianoteq PRO just for the Internal sample rate boost up to 192 kHz from 48 kHz,
I think the main advantage of PRO is the ability to adjust things like hammer hardness, impedance, etc. on a specific range of notes. I'm having some success with hammer hardness and impedance in the top two or three octaves.
However as I have found, it takes a considerable investment of time to be able to make the adjustments, and probably a better understanding of the design parameters and how they interact with each other than some of us have. It's a steep learning curve.
Glenn
]]>olepro wrote:Try to play Pianoteq through a spectrum analyzer and see how much over 10 khz there is ;-)
Exactly. My tests showed almost nothing going on above 16khz. From what I gather, the few scientific studies that have been done on this sort of thing use samples of very bright metallic sounds like cymbal crashes to hear the difference in sample rate quality. Even then, there doesn't seem to be strong evidence that people really can hear differences.
so does that mean that an MP3 is a reasonable format for listening to solo piano work (realizing that recording techniques vary greatly...)?
]]>Try to play Pianoteq through a spectrum analyzer and see how much over 10 khz there is ;-)
Exactly. My tests showed almost nothing going on above 16khz. From what I gather, the few scientific studies that have been done on this sort of thing use samples of very bright metallic sounds like cymbal crashes to hear the difference in sample rate quality. Even then, there doesn't seem to be strong evidence that people really can hear differences.
]]>So what are the conclusions. Perhaps:
1. The end result does not have to be more than 44.1k 16bit (as listening tests typically show), but
2. Use the highest possible sample rates up til the final downsample - to minimize distortions and improve accuracy across the myriad intermediate processing stages (even within the one effect).
In other words:
Use each where they make the most difference, not try to maintain that 'one size fits all'.
It seems (after reading the posts), that if one is going to do considerable audio editing, then the higher rates are beneficial, otherwise not necessary.
My question is - how much are you going to manipulate your wave files from Pianoteq? Personally I do some mild compression (reduce peaks - no attack or release smoothing), and then maximize to -0.3 dB. This is not heavy processing.
Glenn
PS - the audio test or wave vs mp3 will guide you.
]]>Looks like it does have it's uses, although for just listening to Pianoteq, live, perhaps not really.
Greg.
]]>Regarding aspects of the sound that would be improved in going to 192kHz from 48kHz, I suspect that one would need to a) have excellent hearing, and b) listen VERY carefully, in order to hear the slightest difference.
Has anyone noticed any difference at all?
Greg.
The average adult (male) listener can't tell a 16/44.1 wave file from an mp3, let alone something rendered at higher sample rates/frequencies.
Our first year engineering physics prof did an interesting experiment; he brought in an audio sound generator, and started the frequency quite low - 220 maybe. He had us all stand up, and when we could no longer hear the sound as he increased the frequency, we were to sit down.
When the frequency got to 18 kHz there was a handful of students left standing out of an original 150 or so. At 20 kHz there was one guy left, and he looked like he was 14 years old. Remember there is a strong tendency to remain standing after the sound is gone.
Do a test - render a wave, save it and convert to mp3. Have someone else help with a listening test by playing them back - without the test subject knowing which file is being played. I'd be interested in the results.
I'm not suggesting that the musician members of this forum are average - far from it - but who is in your listening audience? And how many of us are still teenagers?
There may be an advantage with higher rates/frequencies in that better equipment will produce the audible ones better, but I'm a skeptic when it comes to numbers like 192.
Glenn
]]>Second to third class...
]]>Regarding aspects of the sound that would be improved in going to 192kHz from 48kHz, I suspect that one would need to a) have bat's or whale's hearing.
Fixed it for you.
]]>