Stereo Tool

I wonder how the block diagram looks like with all these extra filters etc


Is there one?

The sound in this song is pretty insane. When the loud bass (at 50 Hz, so it's not even THAT low) is present, it's 50 (!) dB louder than the mid and high frequencies. That's so insane that I don't think I should even try to fix it... (Especially since dualband AGC handles it pretty well, given the extreme nature of the sound - as I mentioned earlier, 2-band AGC was intended to handle loud bass sounds, which is does!)

On top of that, in v4.22 it sounds bad even with dualband pre-limiter, so (if you use 2 AGC bands) the new version is much much much better already.

No. I might make one when I'm completely finished (things are still changing too much right now). Actually it shouldn't really be that different from v4.22, I've mainly improved existing filters and added filters that remove artifacts that are introduced by other filters (but I probably wouldn't draw those in a block diagram anyway). The biggest change is that Pre Limiter is replaced by an AGC filter.

The order of processing for the most important filters is:
1. Noise Gate (6 bands)
2. AGC (changed)
3. Stereo (5 bands. If 'Remove excessive reverb' is turned off, this is combined with Multiband, and hence 10 bands)
4. Pre-emphasis
5. Lowpass filter (400 Hz above set Lowpass level, only if Lowpass freq > 10 000)
6. Multiband + clipping (10 bands)
7. Highpass and lowpass filter
8. Singleband compressor
9. Loudness: Bass protection
10. Bass boost (new)
11. Final Limiter
12. Loudness
13. Hard Limit
14. De-emphasis and/or MPX encoding

Changes are marked as (changed) or (new), as you can see there aren't that many changes.

(Internally, filters have been changed. For example, the Loudness filter internally consists of 20 different filters, some of which are new...)

i agree .. song is extreme case, do not bother ..

Not only that, but if I would fix it it would change the behavior for other songs as well. Which would probably mean a degradation.

True.

I forgot one thing .. you said that CPU usage will be higher, on my transmiter PC is still same, about 30%.

Ow ok.. I hadn't measured it yet, but given the change I made in the code I had expected it to be higher. If it's not that's of course good news

My next two steps will be:
- Fix ILDP bass protection problem if bass phase differences are present.
- Attempt to lower the Multiband CPU load a bit more.

Here's another copy of the todo list (I had to look back to page 43...)

TODO:
- Remove (small) remaining peaks in Final Limiter - they are a bit higher now due to the improved bass handling ■ OK
- Fix very bad Final Limiter quality in low latency settings. ■ OK, fixed. Equal to v4.22 now, but I think I have some more improvements that I can try...
- Beep sounds are not completely constant in volume when they come out - check why and try to solve it. ■ OK, was caused by Pre Limiter, also Highpass filter
- Fix Stereo Boost - I found that it's behaving very differently from what I expected, but I don't know why. ■ OK
- Fix Singleband compressor. ■ OK
- Fix Stereo Boost meter (shows wrong levels, too low - sound is ok though) ■ OK
- Theoretically the CPU load of BETA3 should be clearly LOWER than v4.22 (a lot of optimizations should reduce the CPU load) - but it's only 2% lower. Check why. (Answer: FM: Oversampling in BETA3A, on 2nd CPU in v4.22. Todo: But what about non-FM? - Answer: Multiband is processed twice in Overlap=0 mode! Impossible to fix. ■ DONE
- Improve Deep Bass Boost for higher frequencies (upto ~ > 100 Hz? Probably with a sliding scale?) See post Bojcha on page 40. ■ DONE
- Fix broken sound when FM filter is turned on without Multiband ■ DONE
- Loudness: Deep bass protection in Overlap=0 mode is less good than in Overlap>0 mode. ■ DONE; checkbox added to enable removing DC component (worse quality!)
- Figure out what the difference is between AGC and what I'm doing now, see viewtopic.php?f=18&t=720&start=20 (among others). ■ DONE
- Add separate processing for high frequencies in AGC filter to fix drop in highs when loud bass kicks in. Simple solution: Lows: Process same as now (total signal, then filter); highs: process same as now, but add a constant value for the lows (first filter, then add constant, then process, then remove constant). Make this configurable: For pumping this should be turned off (single channel should be used.) ■ DONE
- AGC: Add code to limit volume rise for high freqs too much above low freqs -> sounds really annoying and 'thin'. ■ DONE
- Attempt to improve the highpass filter in Overlap=0 mode. ■ PARTIAL, works for 30 Hz phase linear at higher latency values (2048, 4096), there's now also a non-phase linear version for 20, 30 and 40 Hz that causes NO artifacts at all. Question: Should I ALSO offer the possibility to use v4.22 filtering? It's better at high freqs.
- Fix highpass filter: Tests show that bass quality in Overlap=0 mode is better than in Overlap>0 mode. If that's the case, then the highpass filter should also be better. But right now it's not. ■ WAITING FOR FEEDBACK
- Fix Hard Limit in Overlap>0 mode (completely destroys the sound at lower latencies at the moment). Noise gate OK, Stereo OK, AGC OK, Multiband OK, Bandpass OK, Singleband OK, Limiting OK, Loudness OK, FM OK
- Make CPU load configurable in Overlap=0 mode: Noise Gate, Stereo, AGC, Multiband, Bandpass, Singleband, Limiting, Loudness, Hard Limit, FM ■ DONE
- Multiband: Highest band max volume must be 77.5% of that in v4.22 for the same sound, why? ■ DONE, was a bug
- Multiband/Output filter: Check order (compare with v4.22 after fixing the other Trumpet related issues). ■ DONE, moved lowpass filter at 400 Hz above lowpass frequency to before Multiband.
- Restore quality/latency display based on the new settings. ■ Mostly done, todo: check if Overlap>0 should still be supported. ■ NO, bass in Overlap=0 is better. This sometimes hurts mid-highs a bit (vibrating female voices). That should probably be solved by lowering the bass level (or clean/punch slider) a bit. But the difference is not big, and except for this Overlap=0 only has benefits (much lower CPU load and much better bass)
- Move benefits of Overlap>0 mode into Overlap=0 mode by changing the window shapes. ■ DONE, didn't really help (introduced either bass rumbling or higher frequency harmonics).
- Fix problem with ILDP (Loudness) if bass phases are different.
- Hard Limit high CPU load for low J_STEP values. ■ DONE, turned out to be a more generic problem. Current settings seem to give the best performance for all different latency settings.
- AGC causes pre-echos ■ FIXED, less steep filter used
- Improve AGC behavior: Lock high frequency channel to low frequency channel and make sure that its volume never gets lower.
- Improve AGC channel separation at lower latencies.
- Fix Multiband high frequencies cause all-frequency drop.
- Fix dual core IPP FFT
- Create memory manager for better performance ■ NO, bug in Intel FFT calls solved, now seems to be no longer needed
- Fix low latency de-emphasis (differs too much from normal de-emphasis) ■ DONE
- Check the interface for settings that are no longer used (for example "Multiband HQ").
- Disable output displays of filters that are disabled due to too low latency.
- Fix Hard Limit filter for Loudness/Final Limiter ■ DONE
- Attempt to reduce CPU load by simplifying processing: Dualband simpler implementation (4%) ■ DONE, replaced by AGC, Stereo Boost simpler implementation (7%) ■ OK, no longer relevant with faster FFT, noisegate with less bands? (8%), combine Final Limiter with Loudness? (8%) - percentages is what these filters use NOW of the total CPU load in FM Loud Bass Europe mode, replace OvershootBands filter - possible performance gain is a lot lower.
- Check pre-ringing filter effect on Bass Boost (might cause distortion) ■ BUSY, I think it's done but I still hear something, and I need to test it a bit more
- Trumpet filter and distortion in female voices: Create a new Trumpet filter without distortion.
- Fix FM Calibration memory error (and if possible, reduce the latency)
- Create new Low Latency output (original one isn't working anymore).
- Fix Stereo Boost meter (stops after few/many minutes/hours of work)
- Fix non-SSE2 version
- Load and save all the new settings to/from disk

Later:
- Add GATING to compressors (stop increasing volume when the input volume gets below a threshold, reduces pumping and very loud volumes after a period of silence).
- Add "Phase Rotator" usefull for microphone processing
- Add Pre-Emphasize for Normal Output (10,15,20,25us), useful for web streaming (?)
- "Do you really want to close ..." message in stand alone version
- Add "LINK" checkbox for "phase differences" in "Calibrate FM transmitter..." (20Hz-10240Hz), [someone call it TILT]
- Enable separate loading and saving of FM calibration settings. ( viewtopic.php?f=16&t=1370 )

i agree about that. i did't notice any.. but if possible then ok

I actually found 3 bugs that all affect the bass quality:
- Loudness: The left channel had unwanted volume drops if very deep bass sounds were present
- Loudness and Multiband: If the left and right channel deep bass was different, the new filter that removes pre-echos clipped it at far too low levels
- Loudness and Multiband: That filter was also causing some artifacts even if the sound is equal on both channels. (Only in Overlap=0 mode, which is why I didn't notice it when I created the filter). Very loud very low bass sounds sound better now (far less 'rumbling').

I noticed one bad thing: The dualband AGC is causing pre-echos! Will fix that ASAP!


I'll upload a new version tomorrow (it's far too late now to wait for the build to finish)