- Jun 8, 2010
- 1,654
- 1
- 38
SRC | Need to Know
- Thread starter timislegend
- Start date
Joshua Wickman
Yes Sir!
BLUElightCory
Member
It is a bit confusing, but if I'm interpreting it correctly the Pro Tools "sweep" graph denotes good performance, as the swept wave looks to be accurately reproduced without any artifacts. PT is one of the better performers in most of the tests.
guitarguru777
Member
The Izotopes are the best the less cross hatching the better.
Sonar 8.5 beat out Pro Tools thats pretty crazy
Sonar 8.5 beat out Pro Tools thats pretty crazy
BLUElightCory
Member
Yeah, Izotope stuff performed really well.
It doesn't look like Sonar really outperformed PT at anything though (?) A couple tests are close though.
guitarguru777
Member
PhilR
Studio Scapegoat
These tests are pretty misleading. The principle advantage to using higher sampling rates is it pushes digital artefacts and distortion far beyond the frequency limit of human hearing.
Just work on your mixing skills and don't worry about it.
Just work on your mixing skills and don't worry about it.
Jarkko Mattheiszen
The FU guy.
I feel like an idiot, but I really don't understand the graphs. Then again, I don't actually care, either 
I feel like an idiot, but I really don't understand the graphs. Then again, I don't actually care, either
same here
TheDriller
Member
I feel like an idiot, but I really don't understand the graphs. Then again, I don't actually care, either
The main, bright "curve" represents a frequency sweep of a sine wave from 1hz to 20khz. the crosshatch line and other bits of shite represent distortion and artifacts resulting from the Sample-Rate-Conversion being used.
generally, the more stuff on the graph, the less clean the conversion is.
I'd like to see the graph for Ableton Live 8, seeing as the internal SRC engine was reworked after version 7.
Jarkko Mattheiszen
The FU guy.
Yeah, but if I'm just looking blindly at the graphs, I get the impression that Sadie's SRC sucks ass compared to Audacity. Therefore I somehow just can't be bothered with the test results as I have no idea how they translate to practical real world scenarios, if at all.
TheDriller
Member
well the only real-world scenario they apply to is when converting from one sample-rate to another.
Reaper can host files of different sample-rates in the same project, so it applies to any playback situation with multiple sample-rates involved.
Pro-Tools needs to perform a conversion before the audio can be imported, so thats where it applies for Pro-Tools users.
Ableton performs an SRC when using the "warp" feature on audio clips.
Basically, in any of these real-world situations, some DAWs are introducing a lot more aliasing and artifacts than others.
Whether any of it is audible or not is another matter entirely.
Reaper can host files of different sample-rates in the same project, so it applies to any playback situation with multiple sample-rates involved.
Pro-Tools needs to perform a conversion before the audio can be imported, so thats where it applies for Pro-Tools users.
Ableton performs an SRC when using the "warp" feature on audio clips.
Basically, in any of these real-world situations, some DAWs are introducing a lot more aliasing and artifacts than others.
Whether any of it is audible or not is another matter entirely.
Jarkko Mattheiszen
The FU guy.
Exactly
I mean, I can see in the tests that in some cases the whole graph is full of junk, but I still doubt the audible difference would be all that huge.
ahjteam
Anssi Tenhunen
TheDriller
Member
^ I'd like to point out that that's the "regular" quality re-asmpling setting in Live 7.
The engine in Live 8 is completely re-designed, for good reason as we can all plainly see from that graph
The engine in Live 8 is completely re-designed, for good reason as we can all plainly see from that graph
ahjteam
Anssi Tenhunen
AudioPhile777
Mathew Cohen
NSGUITAR
Member
- Oct 26, 2009
- 2,180
- 0
- 36
I don't even pretend to know what this fucking thread is about.
Same.
A little elaboration would be nice.
John_C
formerly Skeksis268
all of the graphs that don't contain any red through to white anywhere except in the upwards sweep and perhaps a tiny amount of reflection above 20kHz are fine, I challenge anyone to pick out distortion at -80dB
John_C
formerly Skeksis268
They resampled a sine sweep from 96 to 44.1 in many different DAWs and plotted the output as time (x) against frequency (y) with volume represented by colour. An ideal resampling algorithm would produce the original sweep perfectly, however in reality they all introduce a degree of frequency distortion (the fuzzy blue background and other lines) and in particular many aliase some of the frequencies above the cuttof point of 44.1 (22.05Khz) back into the audible spectrum (the reflected signal you see on the graph, it's there on the final izotope one, but not the other 3)
