Forum Replies Created
-
Posts
-
Hi guys!
I've been gone a long time — dealing with urgent product development as well as building a house and moving. I've had no idle time to go on facebook for literally 6 months, and the little idle time I had was better spent away from the computer. I'm finally pretty close to being caught up. I do apologize for the troubles you've been having, and appreciate how patient you've been so far!
1) Yes, BaOne will restart automatically when running as a service.
2) Any time BaOne is shut down for any reason, there should be a trace left of some kind.
A shutdown event should be logged in one of several ways. There are many different types of logs.
a. If the service is shut down normally, the entire shutdown procedure is logged in the Startup log.
b. If the service crashes, there should be a crash dump (both a .dmp and .txt file) in the dump\ folder, and an entry in the startup log.
c. If any thread inside BreakawayOne freezes up for an extended period of time, first a warning is issued in dump\BreakawayOneApp_freeze.log.txt, and if it stays frozen for long enough, it repeats the warning and adds “- TERMINATING”, and then terminates the process, so that the service can be restarted. This happened once according to the log files you sent me, while reading streaming metadata, but temporary freezes appear to be happening regularly, in the streaming encoder thread.
That does not make sense to me — according to your settings, metadata is simply being read from a file on your drive. That should not be taking significant time. That is something I need to look into — there could be a bug in my code, or something else could be going on that I’m not yet understanding.
As designed, the only way it should be possible for BreakawayOne to exit without leaving a trace is if the process is forcibly terminated, for example with TASKKILL or with task manager.If Breakaway is crashing at the transmitter site, adding the management package to the studio site would NOT help.
I can add an option to log something every few seconds, but I’m not sure what that would tell us, since it still wouldn’t tell us what happened.
My current framework version is 3.19.77, and I’ve fixed several small bugs in multiple parts of the code since the last time I rebuilt BreakawayOne, and added more detailed logging in some places. Perhaps it’s time I build a new version for you to test, as a first step to tracking down and solving the problem.
Best,
///LeifHi Incontro,
For calibration, you really need a modulation monitor. My own software MpxTool is planned for release next month. Otherwise you can at least adjust tilt by looking at the transmitter's modulation meter and adjusting tilt (while using the square wave test tone in BaOne) for minimum modulation.
BaOne actually doesn't support external RDS encoders. It wasn't worth the hassle and complication for just 1187.5 bits per second of data, better to have BaOne generate the RDS itself with the built in optional RDS encoder.
Best regards,
LeifHello doggy,
This is a non issue. The sidebands are created naturally by the biphase RDS modulation process. Filtering more tightly than this is against RDS spec and degrades the signal. If you compare with other RDS encoders you will find that BaOne is among the cleanest ones.
Best regards,
Leif Claesson[quote author=Milky link=topic=5513.msg19562#msg19562 date=1530495467]
As I understand it, the bass mod is made to the final "speaker" settings after the chain (see Leif's comment on previous page), but I could be wrong. I will seek clarification.
[/quote]I figured "what the heck" :). It's actually part of the main outgoing audio.
///Leif
Buffer size 256 at 192 kHz is equivalent to buffer size 64 at 48 kHz — very, very tiny. 768 is more reasonable indeed, glad it worked out!
///Leif
Hi Timmy,
A sensible replaygain level should definitely be enough, because the whole point of replaygain is to attenuate loud tracks to the point where they're no louder than older, properly mastered tracks.. and as such that should be all the headroom they need, even after declipping.
//Leif
Theodore Roosevelt said "Speak softly and carry a big stick". I say dare to let your audio breathe, at +6 dBr or so :).
You mentioned FM — if you're running 44.1 that means you're using an external stereo generator.. I would strongly urge you to try the internal stereo generator in BaOne, which will require running the MPX output at 176.4k or 192k. It won't reduce latency (in fact the Breakaway Core always runs at 44.1 or 48.0k, and if the i/o is at a higher rate it will up and down-sample automatically.), but it will greatly reduce on-air audio quality.
The MOTU Ultralite is DC coupled, so it's perfect for MPX, but you will of course need to run it at 176.4 or 192.0 to take advantage of that.
The low latency processing very similar but uses IIR filters instead of FIR (so it's not phase linear — phase linearity incurs a lot of delay because the natural behavior (with any analog style filter) is that bass takes longer through the crossovers than treble does, thus you must delay the treble to compensate, so that all frequencies pass with the same group delay, which is the definition of phase linear). Also the low latency processing does not use a clipper or pre-emphasis, it uses a look-ahead limiter instead. So, it won't sound exactly like the air, but it works great for the studio.
///Leif
Hello,
Good choice on the audio interface!
From the screen shot, it looks like you got it right. "Common" is the section is in, but it's actually named the "monitor output". By selecting low latency as you did, you are indeed getting the low latency "fake air" audio through whichever audio channels you assigned as the common monitor output in config.
Whether 256 is low enough depends on the sample rate. If 192k then it's definitely low enough – just 1.33 milliseconds. Multiply that by the (in, process, out), add the algorithm delay (4ms maybe) and any other audio interface I/o delay, and you're still under 10ms.
Best,
Leif ClaessonIndeed I check for it so that it doesn't try to use it as VST, because it crashed otherwise :).
BreakawayOne uses 24-bit samples with 8-bit headroom (so levels internally can go to +48dBFS, not limited to 0dBFS).
VSTs are 32-bit floating point, which means 24-bit resolution with essentially infinite headroom.
So, if you feed a 0dBFS clipped modern loudness-war master into BaOne without attenuation, and then run a VST de-clipper, you will have +12dB or maybe even +18dB peaks, which is perfectly fine because BaOne has the headroom for it, and the AGC will push it down to where it should be.Winamp DSP plug-ins, on the other hand, use 16-bit samples and has NO headroom. So, if you wanted to do a de-clipper as a Winamp DSP plug-in, you would need 12 or maybe even 18dB attenuation to avoid clipping in the DSP chain.
18dB.. that's 3 bits. So you'd be down to *13 bits* of audio data!
That's just not enough to be useful, so I skipped it completely.
Best regards,
Leif ClaessonListening to it, I think it mostly sounds upper-mid heavy.
One thing to note: LEIFF does not at all mean Leif's recommended or favorite preset.. Quite the contrary, I can't really stand it.. it's meant to emulate the sound of the GREGG preset in an Optimod. Greg -> GREGG, Leif -> LEIFF, get it? :). It's supposed to be strident and in your face because the original preset is.
Are you using any voice processing? It usually works wonders to disable it completely. All you really need is pre-amp, and perhaps a bit of EQ. If you could apply a single band of parametric EQ to just the voice, let's say 3 KHz 1.5 Octaves -3.0dB, I think you'll be there.
I don't know anything about EMET — what is it BreakawayOne does that wouldn't be compatible? It's a standard C++ program using only official APIs, if it doesn't work I don't know what apps would?
///Leif
Eventually, yes. We're writing one. But, if a manual is absolutely critical for you, I'd hold off on purchasing until it's actually out.
Best,
///Leif
