All times are UTC+01:00




Post new topic  Reply to topic  [ 13 posts ]  Go to page 1 2 Next
Author Message
 Post subject: Stereo Tool 9.36 BETA
PostPosted: Sun Oct 20, 2019 5:27 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11185
Latest version - Beta 011 (Oct 26 2019)

The BIMP Virtual Console
The BIMP now comes with a virtual console, in which you can control the insertion level of all (for now at most 2) microphone inputs after processing. This replaces part of your hardware mixing console! With this it's now possible to have microphone processing, mixing and final processing with only 6 ms latency in total (for headphone output)! (For now it's only built-in, we will make a web interface to run on a tablet later)

MicroMPX V3
We recently discovered that in MicroMPX V1 and V2, when audio started after a period of pure silence, a crackling sound could be audible. This was caused by a bug. (It only happened sometimes, and only after a period of pure silence, which is why nobody seems to have discovered this earlier). The new encoders can encode versions 1, 2 and 3, and the new decoders can decode all versions. If you update, update the decoders before you set the encoder to encode V3 - the old decoders cannot decode V3. If you are using an Omnia MPX Node or Omnia.9 with the latest software version, they both only encode V3, so you need a new decoder.

Downloads
Changes/bugfixes
  • PhoneBooster remove horizontal lines
  • Test Linux sync
  • Disabled PhoneBooster in all versions except stand alone product for now
  • Set PhoneBooster stand alone to only 48 kHz, 44.1 will follow later (and then we can add it to all versions).
  • Include classical music presets
  • Check other new presets

TODO's
  • Fix Declipper build failure
  • Fix PhoneBooster build failure
  • Fix O11 RDS build failure
  • Fix AGC SB settings warning
  • RELEASE
  • Update uMPX Pi image
  • Re-run Kantar test
    -
  • If still glitches, try setting and IMMEDIATELY resetting event.
  • STX Watchdog
  • Finish MicroMPX web interface
  • Make AGC new compressor mode instead of SJP
  • STX screen updates? Make smarter drawing
  • STX fix ASIO drawing at bigger block sizes
  • STX try drawing only part of lines when possible
  • STX RC
  • STX resamplers?
  • PhoneBooster new model
  • PhoneBooster new Normalize
    -
  • Measure HifiBerry stereo separation -> Improved from 35 to 55 dB. Make RC work with Ignore high frequencies -> Later
  • Clipper: Add sensitivity slope for highs filter.
  • Figure out how to dynamically create new GUI structure BUSY
  • STX self test on multiple units
  • PhoneBooster add 44.1 support
  • PhoneBooster above/below 6 kHz?
  • Button 1x on error
  • Shift in uMPX
  • Try to fix Window (only use in first stage, adjust output levels in subsequent stages)
  • Subsidie, pensioen
  • Classical music preset THU
  • How to synchronize HD? THU
  • Test Framework? BUSY
    -
  • AGC classical music settings
  • Loudness project
  • BIMP: Add panning, metering, main audio level; adjust displayed levels? (+6 or 0?)
  • Check scheduler, reported to be partially broken in 9.35 (see forum posts)
  • Complaint: All (Normal, LL) outputs in Linux go to FM
  • Release PhoneBooster beta
  • Reduce CPU load as tried at IBC
  • Tom aan de Stegge remarks email 30 september 20:23 - change many defaults
    -
  • Support rewindowing with different Quality/Size settings. Will give better HPF performance and make lower CPU load possible, and PPM patch points.
  • Windows image
  • Reverb fix weird sounds on Mac. Several possible causes fixed, waiting for feedback.
  • Finish 9 - final uMPX file name issue -> In next 9 update.
  • Figure out how to display NMEA/pulse status in GUI. -> Waveform and jitter-view for pulse detection in detail view, GPS in stream, NMEA data, NMEA readable, Pulse detected flags.
  • PhoneBooster make multi-threaded if needed (it probably is, esp. for Pi)
  • PhoneBooster make Pi version
  • Add something to get now playing content from file (but: first figure out how, could be via a separate program or something).
  • Streaming output can hang - calls probably need to be moved to a separate thread. ???
  • MicroMPX needs exception handling, just in case.
  • WatchCat replace files instead of placing them elsewhere. For playout systems.
  • WatchCat add CMD /C
  • Thimeo School
  • ABDP remove stuff that's no longer needed? BUSY for low CPU
  • Optimize ABDP
  • Input and output gain for command line version (like for sound card)
  • Dynamic adjustment of priorities
  • Sooth effect
  • Crash ProppFrexx
  • Update Omnia.SST presets
  • AES67: Add WheatNet
  • Create file when input fails
  • Installer: Make installer aware of programs like MediaMonkey, AIMP, .....
  • HD delay
  • Document JSON interface
  • Documentation: "Processing" page contains all kinds of BS412 info - why?
  • Declipper L/R noise
  • Declipper handle resampled audio better (upsample, find oversampling spikes, then detect based on what remains)
  • Console version: Read sample rate and bit size from WAV header
  • # bands selections should allow option to insert a band instead. Or ask what to do when you change it.
  • AES67 thread affinities and priorities
  • Add resampler

OLD VERSIONS
Quote:
Latest version - Beta 010 (Oct 25 2019)

Downloads
Changes/bugfixes
  • PhoneBooster find data for frequencies 7000-8000 Hz - Bad idea
  • PhoneBooster make filter settings for 3 sizes!
  • Disable PhoneBooster on bad settings
  • PhoneBooster set fixed quality, latency, samplerate, threads, more?
  • MicroMPX decoder mu character
  • MicroMPX stand alone encoder V3 web interface issue -> V3 disabled for now.
  • Verify preset saving code on Linux - code looks fine, test on Linux to figure out what happens. LINUX DELAYS SAVES. "sync" COMMAND. Scary stuff... To be tested.
  • Optimize ITU1770 calculation (192 kHz -> 48)
Quote:
Beta 008 (Oct 20 2019)

Downloads
Changes/bugfixes
  • BIMP virtual console added!
  • MicroMPX 3 (same version as in Omnia MPX Nodes)
  • MicroMPX decoder GPS precision delay slider added
  • MicroMPX decoder GPS lock display added
  • LUFS values display added
  • Added auto-phase-inversion with detection.
  • Bug fix: Fixed potential glitch in Normal Output when using multiple threads, mainly noticeable when playing input to output.
  • Added sound card self test glitch detection
  • Kantar log file name: Added 1900 to year
  • Kantar log files: Make sure directory is always created
  • Kantar Passed checking procedure
  • Linux image
  • Discuss web interface
  • PhoneBooster 48kHz
  • PhoneBooster check filtering - something is broken with the latest build
  • PhoneBooster compare with previous (IBC) version
  • PhoneBooster optimize THU
  • PhoneBooster change shape or HIYF filtering


Top
   
PostPosted: Tue Oct 22, 2019 3:52 pm 
User avatar

Joined: Sun Dec 23, 2018 7:44 pm
Posts: 792
Location: Texas, USA
Quote:
LUFS values display added
Attachment:
st-1770.png
st-1770.png [ 3.06 KiB | Viewed 8142 times ]
Is this it on the primary processing page? Am I supposed to be seeing number values? It also appears to make audio stutter.


Top
   
PostPosted: Tue Oct 22, 2019 8:11 pm 
User avatar

Joined: Tue Mar 17, 2009 2:56 pm
Posts: 4151
Quote:
Quote:
LUFS values display added
st-1770.png

Is this it on the primary processing page? Am I supposed to be seeing number values? It also appears to make audio stutter.
Yea, numbers needed. Also it can easy be in peek meters. Looks like it's 3sec constant implementation, but it's so steady sometimes, like it's 10sec. Probably because same reason as when there is no audio meter never drops.

_________________
control point
control point2


Top
   
PostPosted: Thu Oct 24, 2019 2:18 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11185
That ITU1770 meter will get numbers, we're working on a new GUI that will be way more flexible. I've added this feature already but without the numbers so far... they will come. If you want to know what you're seeing: All the way to the right is 0 dB, every thick line is -6, so the display in the screenshot above indicates about -7 dB.

It's normal that it doesn't go down when there's silence, that's how ITU1770 works. Very low level audio is ignored in the measurement. That's also what you want, if you're broadcasting a movie for example and there's a few minutes of (near)-silence you don't want to have to boost the rest of the audio to reach the correct target level.

Glitches are probably caused by the CPU usage, I'll try to optimize it (that's why it's turned off by default). One way to optimize it would be to update the displayed level less often.


Top
   
PostPosted: Thu Oct 24, 2019 2:56 pm 
User avatar

Joined: Tue Mar 17, 2009 2:56 pm
Posts: 4151
Quote:
It's normal that it doesn't go down when there's silence, that's how ITU1770 works. Very low level audio is ignored in the measurement. That's also what you want, if you're broadcasting a movie for example and there's a few minutes of (near)-silence you don't want to have to boost the rest of the audio to reach the correct target level.
Ah, ok, then every other loudness meter is wrong.

_________________
control point
control point2


Top
   
PostPosted: Thu Oct 24, 2019 3:03 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11185
Quote:
Quote:
It's normal that it doesn't go down when there's silence, that's how ITU1770 works. Very low level audio is ignored in the measurement. That's also what you want, if you're broadcasting a movie for example and there's a few minutes of (near)-silence you don't want to have to boost the rest of the audio to reach the correct target level.
Ah, ok, then every other loudness meter is wrong.
ITU1770 ignores anything that's below -70 dB, and anything that's more than 10 dB below its average level. Not sure what other loudness meter you're comparing with?

From https://www.itu.int/dms_pubrec/itu-r/re ... !PDF-E.pdf :
Quote:
The algorithm consists of four stages
– “K” frequency weighting;
– mean square calculation for each channel;
– channel-weighted summation (surround channels have larger weights, and the LFE channel
is excluded);
– gating of 400 ms blocks (overlapping by 75%), where two thresholds are used:
– the first at −70 LKFS;
– the second at −10 dB relative to the level measured after application of the first threshold.


Top
   
PostPosted: Thu Oct 24, 2019 3:14 pm 
User avatar

Joined: Sun Dec 23, 2018 7:44 pm
Posts: 792
Location: Texas, USA
Quote:
If you want to know what you're seeing: All the way to the right is 0 dB, every thick line is -6, so the display in the screenshot above indicates about -7 dB.

...

Glitches are probably caused by the CPU usage, I'll try to optimize it (that's why it's turned off by default). One way to optimize it would be to update the displayed level less often.
Thanks, that helps.

A little more info: Not sure what all affects the update rate, but I'm using 150% quality on 3 hyperthreaded cores, at a 144hz refresh rate.


Top
   
PostPosted: Thu Oct 24, 2019 3:19 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11185
Quote:
Quote:
If you want to know what you're seeing: All the way to the right is 0 dB, every thick line is -6, so the display in the screenshot above indicates about -7 dB.

...

Glitches are probably caused by the CPU usage, I'll try to optimize it (that's why it's turned off by default). One way to optimize it would be to update the displayed level less often.
Thanks, that helps.

A little more info: Not sure what all affects the update rate, but I'm using 150% quality on 3 hyperthreaded cores, at a 144hz refresh rate.
The level can be calculated every 100 ms. Because then there's data from a new 400 ms block with 75% overlap. So, that means that a pretty heavy calculation (look at all previous audio, drop everything that's more than 10 dB below the average, and take the average of what remains) is done every 100 ms. That might be a bit much... Enough to cause a glitch every 100 ms, possibly.

Edit: Reduced update speed to once per second (which is fast enough), doing parts of the calculation on the fly so we don't have to examine everything anymore, at high sample rates (> 48 kHz) reducing it by skipping samples before sending it to the (heavy) resampler. Result: Values are still the same but the effect is almost invisible in the CPU load, even in debug mode :)

With 192 kHz input,
Old release CPU load with measurement disabled: 0.1%
Old release CPU load with measurement enabled: 4.2%
New debug CPU load with measurement disabled: 1.5%
New debug CPU load with measurement enabled: 2.2%


Top
   
PostPosted: Fri Oct 25, 2019 10:44 am 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11185
BETA010 is available!


Top
   
PostPosted: Fri Oct 25, 2019 12:41 pm 
User avatar

Joined: Tue Mar 17, 2009 2:56 pm
Posts: 4151
Windows10:
Publisher: Unknown publisher.

_________________
control point
control point2


Top
   
Display posts from previous:  Sort by  
Post new topic  Reply to topic  [ 13 posts ]  Go to page 1 2 Next

All times are UTC+01:00


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Limited