RE: Earlier discussion about use of Hi-Q mode affecting gain

[nms]

So earlier I mentioned I have been avoiding using Hi-Q mode in the clip settings because I was sketched out a bit before about how much it seemed to alter the sound just because I expected a high qual setting to be more transparent if anything. I just loaded something up to try it and measure.. and it causes a 1.04 db increase if I use the setting! So my audio clip that was normalized previously now clips the meter an extra db. I used a 24bit 44.1khz audio clip for this test. my soundcard's set at 48khz.


Robert Henke.. you mentioned this:
"The reason why the hi-quality SRC can slightly go over 0.0dB is in the nature of the interpolation. But seriously, who cares????? SRCs need to create sample where no sample was before. If this leads to a sample with the value of 1.0000000000000001 at some point, and this is the best possible interpolation value to get a smooth curve, why bother?"

but 1db seems fairly significant.

[SubFunk]

Move the master down 1.04 db... Does it sound better to YOU or not?

[Robert Henke]

1.04 dB is a lot, but the question is, how exactly looks the resampled file? is it a single transient that pops out for one or two samples, is it a low frequency part that shots over for a longer period of time?
I would be curious to see the waveform of the portion that peaks and compare it with the waveform before the SRC.
Robert

[nms]

here's a pic.. though this is easy enough for anyone to replicate.:



this is 10 seconds of a song, and the clip marks are 2 sample clips. This one actually peaks at +1.62db
in 10 secs of audio there was only one clip which was 3 samples long. It's mainly 1 & 2 samp spikes so I take that as it's boosting & clipping the high end, which makes sense because I mainly just notice a brighter sound when the HiQ button is engaged.

[nms]

so what's going on here, is this just an illusion of higher quality brought about by a brighter signal?

[nms]

alright I've figured out that this increase when toggling the Hi-Q clip mode only takes place on files that are different than ableton's selected sample rate setting. This was my case, as my soundcard was set to 48khz and the file was a 44khz wave. Not sure why this gain alteration only takes place with Hi-Q engaged though.

[IP]

Some strange things happening when working with files with deferent SampleRate than the projects SR
(i mean the default SR you have set to your interface/live settings).

In my opinion if you choose to work @ 44.1 (because your most files r 44.1 or for any other reason)...
everything have to be 44.1

The non 44.1 samples have to be converted to 44.1 before they imported to project
The settings have to be @ 44.1 etc.

This is what i do and what feels right to me

[ethios4]

alright I've figured out that this increase when toggling the Hi-Q clip mode only takes place on files that are different than ableton's selected sample rate setting. This was my case, as my soundcard was set to 48khz and the file was a 44khz wave. Not sure why this gain alteration only takes place with Hi-Q engaged though.

Well yea, the Hi-Q setting is referring to high-quality sample rate conversion, which only takes place if there is a difference between the file's sample rate and Abletons. If the sample rates are the same there is no conversion and thus HiQ does nothing.

@IP - It's not strange that where is a difference when you are working with different sample rates than Ableton's sample rate. Sample rate conversion will always create a difference - you are creating new samples in the file! The process you describe (converting samples to the desired sample rate outside of Ableton) is exactly what is recommended in the manual. Live's SRC is designed for live performance situations, and while very good in HiQ, is not quite as good as the best offline SRC. Again, all this is spelled out in the Ableton manual.