Stereo Tool

Those were very long hours...

Anyway, the new version is online now.

Changes:
- Pre-emphasis without latency
- De-emphasis without latency
- Multiband compression: Fix so highest bands are better separated
- Processing blocks reduced from 1152 to 64 samples in low latency mode
- Low latency mode uses only 1 core (!)
- ASIO-latency can be reduced to 2 ms.

New bugs:
- Changing from low latency to normal more can cause hangups
- When using deemphasis, some frequencies may end up louder than they were - which could lead to distortion in the normal (non-FM) output.

wow...
i am geting 20ms exactly... buffer not working at 0.002 but works at 0.003 with my e-mu0202.
Now we can call it real-time processing.. True!

Another question ...
In LowLatency mode .. i hear some "white noise" ? in previous beta too. Can that be fixed ?
it's something with NoiseGate, like not filtering highs.

Ok - I still need to check what's causing the additional latency.

About the noise gate: I hadn't tested it that thoroughly yet, but I was afraid something like that could be the case. The problem is that the channel separation is very poor. I'll do some tests to see if I can fix it (for example by _only_ filtering the high freqs), if not I'll replace it with a very-low (but not zero) latency step with only 2 or 3 bands.

- Edit: I just realized that this filter cannot work at all. If there's some bass content, the highs will get some signal too and no filtering will take place. So I'll need to use the same filter I use in the normal latency version - except that limiting it to 2-3 bands will greatly reduce the latency that's required.

i don't belive much to my e-mu 0202..
tomorrow i'll try with maya44.

Have you tried it with 44/48 kHz input? At higher sample rates Stereo Tool needs to downsample and upsample it, which causes extra latency.

Note: I think I just though of a new clipping method that might work with only a few ms latency, and won't sound worse than the current clipping (only for higher frequencies)

i have 20ms with 48kHz input
multiband, noisegate, singlecompressor and Hardlimit. hq=on

Ow, that explains a lot - I haven't looked at Hard Limit yet, and currently it causes a latency of 11 ms. I *think* I can reduce that to about 3 ms without too much effect on the sound...

Note: Noise Gate is fixed now (not yet uploaded), causes an extra delay of 1.5 ms. It only works properly on frequencies above 4000 Hz.

ohhh.. true
i disabled Hard limit .. overall latency is 6-7ms

Latency of Hard Limit is now down to 1.1 ms, with very little effect on the sound as long as there are no loud bass peaks (and they SHOULD have already been removed before the Hard Limit filter kicks in, because it's intended to remove small peaks that remain after the rest of the processing only).

Noise Gate has a latency of 2.9 ms. So that's 4.0 ms in total on top of those 6-7 ms --> ~ 10-11 ms.

I expect that the multiband clipper latency will be 3, 6 or 12 ms. (6 seems to sound ok, I'll try if I can reach 3 without introducing too many artifacts. It won't be able to clip low frequencies).

Unfortunately there's a lot more that I need to add:
- Lowpass filter
- Probably a highpass filter also to remove frequences below 30 Hz
- Final Limiter
- Loudness filter
- Upsampling and downsampling (44 / 192 / 572 kHz)
So I fear that eventually the latency will go up a lot more. Having said that, I've been able to greatly reduce the latency of every filter I've looked into so far...

New version is available now - same links as before!