Ok, I have a 24-bit mono file loaded into a track and removed the vol/pan filter to avoid the internal mixer as I will be routing to outboard.
My question is: does T2 convert every 16/24-bit file into 32-bit internally; will my ASIO driver see a 32-bit signal or just the 24-bit one? I can choose from multiple precision ASIO drivers, so I feel I need to know for optimal performance.
T2 tech question
-
- KVRAF
- 4644 posts since 28 Nov, 2002 from Chicago
My understanding is that Tracktion mixes entirely at 32bit - so if you are using 24bit drivers, recorded audio will be converted up to 32bit and then back down to 24bit.
Given that the conversion between 24bit FP and 32bit FP is trivial though, this isn't going to affect performance to any meaningful level.
Given that the conversion between 24bit FP and 32bit FP is trivial though, this isn't going to affect performance to any meaningful level.
Someone shot the food. Remember: don't shoot food!
-
- KVRist
- Topic Starter
- 179 posts since 15 Oct, 2003 from Norway
I am not sure the conversion is that trivial given that 24bit is actually integer format while 32bit is float. Even though 32bit FP in reality is 24 bit precision with an 8 bit scalar.
My question would then be, does T2 convert back to 24 bit after mix to match the ASIO drivers preference? And what if the signal never reaches the mixing engine (not routed to "default audio output)? In any regard I will not use the internal mixer at all for this task.
Thank you for your input.
My question would then be, does T2 convert back to 24 bit after mix to match the ASIO drivers preference? And what if the signal never reaches the mixing engine (not routed to "default audio output)? In any regard I will not use the internal mixer at all for this task.
Thank you for your input.
-
- KVRAF
- 4644 posts since 28 Nov, 2002 from Chicago
24bit is Int? Are you sure? I'd have to go back and read the ASIO spec to be certain myself, but I was under the impression that 24bit was FP.I am not sure the conversion is that trivial given that 24bit is actually integer format while 32bit is float. Even though 32bit FP in reality is 24 bit precision with an 8 bit scalar.
Signal is *always* passing through an internal summing stage somewhere along the line.And what if the signal never reaches the mixing engine (not routed to "default audio output)? In any regard I will not use the internal mixer at all for this task.
The moment audio is received by Tracktion it is going to be in the domain of T's mixing engine. The different audio outs are beside the issue. You can send many tracks to any given audio out, therfore clearly Tracktion is mixing them.
Someone shot the food. Remember: don't shoot food!
-
- KVRAF
- 6740 posts since 25 Mar, 2002 from sheffield, england
-
- KVRist
- Topic Starter
- 179 posts since 15 Oct, 2003 from Norway
Yes, I am pretty sure 24-bit is integer format, at least the files are.valley wrote:24bit is Int? Are you sure? I'd have to go back and read the ASIO spec to be certain myself, but I was under the impression that 24bit was FP.I am not sure the conversion is that trivial given that 24bit is actually integer format while 32bit is float. Even though 32bit FP in reality is 24 bit precision with an 8 bit scalar.
Signal is *always* passing through an internal summing stage somewhere along the line.And what if the signal never reaches the mixing engine (not routed to "default audio output)? In any regard I will not use the internal mixer at all for this task.
The moment audio is received by Tracktion it is going to be in the domain of T's mixing engine. The different audio outs are beside the issue. You can send many tracks to any given audio out, therfore clearly Tracktion is mixing them.
You do get 32-bit integer format aswell, but it is not used that often.
Hmm, I do believe tracks only sent to "default output" are sent to the mixing engine. If you have multiple outs, try sending some tracks to any output that is not set as the default output. If they do not show up in the master then I doubt they ever reach the mixing engine.
Perhaps I should enable normalizing or something.
-
- KVRAF
- 4644 posts since 28 Nov, 2002 from Chicago
The WAV format is integer by default.voidar wrote:
Yes, I am pretty sure 24-bit is integer format, at least the files are.
The VST spec is entirely FP. I haven't looked at the ASIO SDK in about 4 years now, but I recall that also being FP (which given the FP state of VST would make sense).
That said: JUCE which is the framework behind Tracktion is FP native, so I suspect Jules converts everything to FP as a matter of course.
The master out is not the mixing engine. It is just a convenience path for the user. Consder a standard mixer. There is a master mixer strip and sub output strips. Your extra audio outs can be thought of like this.Hmm, I do believe tracks only sent to "default output" are sent to the mixing engine.
If you can send more than one track to an audio output, and you can apply any effects to those tracks, which you can, then logically they *have* to be going through a summing engine.If you have multiple outs, try sending some tracks to any output that is not set as the default output. If they do not show up in the master then I doubt they ever reach the mixing engine.
Someone shot the food. Remember: don't shoot food!
-
- KVRist
- Topic Starter
- 179 posts since 15 Oct, 2003 from Norway
A 24-bit file is a 24-bit file just as a 32-bit float file is written as floats.
If you put it that way, then yes, those tracks must be summed. But they are vectors added together, and thus I would presume they are either float/double precision with values between -1 and 1. And because of that I would guess a 24-bit file would be converted prior to the mixing engine.
So I guess that answers my question. I think I should turn dithering on and then I would be fine.
As long as I record in 24-bit I guess it is a choice between the recorded files integrity vs. the integrity of my monitoring.
If you put it that way, then yes, those tracks must be summed. But they are vectors added together, and thus I would presume they are either float/double precision with values between -1 and 1. And because of that I would guess a 24-bit file would be converted prior to the mixing engine.
So I guess that answers my question. I think I should turn dithering on and then I would be fine.
As long as I record in 24-bit I guess it is a choice between the recorded files integrity vs. the integrity of my monitoring.