Voice alerts test build

Moeder wrote: @Morlacus: attached is a sample set of french mp3s and corresponding map file for you to enjoy

Very kind of you. Thanks a lot

Some more testing with my devo 8s.
The welcome message replaced by the RUDD DR position message occurs when the throttle stick is set from -43 % to 45 %. I is not replicable on the emulator . I would be interested to know if anybody has encountered the same weird thing.

With my TX started, I changed the model to a ppm protocol one in order to play with a simulator. All worked well. I shut down the TX. When I turned it on again, I had only a flickering splash screen. A long press on the ENT button cured the TX and I was able to access the model and change it.

I would be interested to test the build on the devo 12S but I would begin by the emulator as I think that Aux knobs apply to this model and I do not see how to have music on them. But there is no devo 12 emulator.Shoud I take devo 12 e one ?

I will look into the PPM problem and rethink your RUDD DR position message. I uploaded the Devo12S Win Emulator in to the test build page to play around with. The current AUX implementation is the one by phantom-8, which basically has two sounds selectable for AUX in positive and negative position. I was actually thinking about additionally announcing the current value (from -100 to +100) of the aux knob.

Edit: this reminds me I didn't take into account negative telemetry values (perhaps from temperature sensors?) in the code. I will be a short fix, but while I'm at changing music map again, is there any other preset mp3s we would possibly need? We might just add minutes and hours, in case we need them later on? Or should we take the complete set from opentx so we're set for all future extensions(milliamps, kmh, mps, mph, knots, meter, foot, celsius, Fahrenheit, percent, Watt, dB, ml, founce)?

Please check if the PPM bug has been resolved.

I updated the builds, F12E is also now supported.

Moeder wrote: Please check if the PPM bug has been resolved.
]

Unfortunately no change as concerns the ppm bug . Always the flickering splash screen (I forgot to say that some times the press on enter triggers the USB screen)

As concerns units, in addition to those already in the music.map I agree on foot, fahrenheit which are already in the tx config. I hope minutes and why not hours for a new countdown possibility. Percent if you think about giving a detailed message for Aux knobs. All units possible in telemetry messages should be there also. I suppose km/h mph, Watt. I have no info on telemetry giving info in ml or ounces or DB, Milliamps are treated as point xxx amps. And I imagin that g which is in the present music map will not be very usefull, but ...

I replaced devo8 build with a developer build which has debug log menu. After starting your TX with the flickering screen, once you regained menu access go into debuglog menu and post what you can find there.

Moeder wrote: I replaced devo8 build with a developer build which has debug log menu. After starting your TX with the flickering screen, once you regained menu access go into debuglog menu and post what you can find there.

I tested the debuglog build.
I had no flickering after the ppm protocol model.ini. And after reloading my test model with sound I had no sound.
I enclose a snapshot of the log . Is there another way to record it ?
To check I came back to the original 5f40044.dfu. I had again flickering and sound.
Sorry for these results which seem somewhat inconsistant.

So it seems to have something to do with the handling of ppm detection in the audio code. I will look into it.

Edit: could you also post debug log of loading ppm model and then switching to sound model?

Here are some more pics of the tx back to sound model. I did not hear the sounds mentionnned in the debub log.

Thanks, I went through the code and noticed audio output commands are not enabled in development builds
The current way of setting up the port and detection of its use by ppm is rather crude, I will change this hopefully solving the problem. Any other bigger problems noticed? How do the aux knobs work?

Happy if these test helped you. My main remaining problem is the Rudd DR message in place of welcome as mentionned.

As concerns Aux knobs on devo 12 I have just seen how they work in the emulator. I have never used them as swiches and when I use them I use them as proportional channels so no real interest to have a sound for that. I have not tested any build on my devo 12 because it is more complex to install and uninstall (make the library.dfu with my media file). And I use my devo 12 for flying.
My main wish on the voice system is a count of elapsed time in minutes and hours but I have written a lot about this...

Concerning Aux: that is why I was thinking it might be useful to have numerical output on them

Concerning Timers: I hear your wish. I'm actually thinking of making timer alerts configurable the same way as telemetry alerts, it will be only 4 additional entries in music menu and will enable you to select a custom sound for each one. Since you might use countdown and stopwatch timers differently on some models, the preset voice alert (currently "Timer X alert") has to be pretty general. I would then add numerical readout in a custom interval for stopwatch timers as well. We will just have to think of the right place to set up timer announcement intervals, or should we use current Timeup interval for starters?

Moeder wrote: Concerning Aux: that is why I was thinking it might be useful to have numerical output on them

May be I have not correctly understood your intention but this seems difficult to do.if the slider or the knob is moved frequently this could activate numerous voice messages. The message should come only when the aux channel is inactivated for a certain time.

Moeder wrote: Concerning Timers: I hear your wish. I'm actually thinking of making timer alerts configurable the same way as telemetry alerts, it will be only 4 additional entries in music menu and will enable you to select a custom sound for each one. Since you might use countdown and stopwatch timers differently on some models, the preset voice alert (currently "Timer X alert") has to be pretty general. I would then add numerical readout in a custom interval for stopwatch timers as well. We will just have to think of the right place to set up timer announcement intervals, or should we use current Timeup interval for starters?

As concerns timers I think the count down is good. the interval can not be less than 10 seconds taken into account the duration of the message. "time remaining"+ "x"+ "seconds" . That's why I imagined to have the message before the time and a beep on the real second. May be this is really too complex.
Prealert interval and time up interval are good for me. but an "elapsed time" should be put in range of minutes (say one to ten) or set automatically in fonction of the elapsed time (for count down a box to fix the interval and for stop watch for example 1 minute until 5, 2 up to 10 an 5 after). the good place for me to put this interval would be in the model timer menu for each timer.

I installed the deviation-devo12e-v5.0.0-5f40044 build on my 12e. The voice works great and I love being able to set the volume. I have configured all the switches, which appear to work very well.

The sounds play well. They do run sequentially when the queue is full and it seems like it takes a while to play them all sometimes. Perhaps we can consider doing sound compression on the files so that the play faster; I can try to work a few here. There are minor issues:
- the MIX sounds actually spell out M-I-X-0, M-I-X-1 and M-I-X-2
- the AUX sound for AUX 7 (the only one I tested) played multiple times when I turn it down or up. It played the right sound (e.g. "flight mode deactivated" when turning down) several times, which would certainly fill the queue for long sounds.
- No welcome or shutdown sounds.
- I haven't got anything set up with telemetry at this point, but I'll test as soon as i can.

I've been working with a new circuit board and enclosure for the external Voice Box. I have combines the dfplayer, a micro-usb jack, 3.5mm mono jack, busy-indicator and up/down volume buttons, as well as a 29mm speaker. This is built into a 3D-printed box, along with a 200-300mah battery. It isn't quite right yet... I am using a 60x40mm pc board (common size) that doesn't quite fit, and the access to the dfplayer SD card is very poor (although you can still access the files via USB). Overall size is about 66mm x 45mm x 25mm. I'll post a photo or video when I get a chance.

I do get welcome sounds. I extended the splash time from 35 to 45 and now I get the sounds. I get both the "Flight Mode Activated" then "Welcome to Deviation" sounds.

I am not getting any sounds on shutdown.

I tinkered with FDR's new cleaned-up splash screen. It is supposed to be a megaphone... Devo with sound !

Tested on my 12 and all seems to be working well accept the GoodBye which to me is no big deal..

Have been testing test build from 22 Feb 17 on my U7E, also soldered a wire to PC12 so not longer use the PPM pin and can confirm that this also works. (just edit hardware .ini)
I have to say thanks so far Moeder, getting better and better! Some points from mu side;

Minor things:
-Welcome to deviation, in my case only sometimes it is being played, most of time not. (not clear excactley why it happens)
-Goodbye at shutdown never played.
-Volume, possibly discard/empty the queue of played messages and only play last one. (if scroll fast you get volume, volume, volume volume etc)

Music menu:
-For switches pots, maybe the fields to be consistent/similar to the other deviation setup menus. So use Src detect, when flipping the switch moving the pot, the particular switch, stick, pot etc is detected and filled in the field.
- Possibility to setup a switch to put some multiple actions: F.I. when flip switch speak some assigned alerts; Battery voltage - Flight time - RSSI
- How to setup a spoken numeric value or, just play an alarm. F.i. TX battery low, we could play just a sound "horn.mp3 , or we could have message <warning> <TX battery> <4> <point> <4> <Volt>

- Repetition rate for alerts ?
- Threshold for alerts, only start play the alert, when threshold made for set time in ms (not really a voice option but more general, important to avoid many alerts, driving you crazy)

Thanks for all the much needed feedback!
I've just uploaded updated builds with many improvements:
1.) PPM usage detection was completely rehauled, I sincerely hope the flickering screen bug is gone. In rare cases where you turn off PPM usage in a model file you will have to reload the model (or faster turn on/off TX) for reenabling audio for the specific model.
2.) Timer alerts have improved timing, the message will start playing earlier so the number mp3 will match the timer time in that specific moment (+ a couple ms of dfplayer delay).
3.) Aux channels will now trigger sound when they are turned and left at new setting for 1 second. They will also announce their value. I didn't have time to set up a new music.map yet, so the "minus" for negative aux-knob values (as well as telemetry values) is omitted for now, but will be there in the future.

I'll try to answer to your feedback:
-Goodbye message: Haven't looked into it yet, it seems too unimportant
-Volume announcement: I've found this annoying in the emulator as well, I'll change it to only announcing volume when the queue is empty as this seems to be the simplest solution
-Src detection in music menu: this was my first solution, but I dropped it due to many reasons, some of them we discussed in this thread.
-additional types of actions for music and selecting between spoken numeric or just alarm: this will mean quite some more coding, and honestly I won't have the time to add these in the next days or weeks. And we have to leave at least some room for future improvements
-repetition rate for alerts: for timers and battery they are set in transmitter config, for telemetry they are currently hardcoded to 15 seconds. We might move this to be configured in transmitter config as well
-MIX sounds: it's a matter of how the TTS solution handles it...feel free to realize your very own sound set. I'm sure once this is included in nightlies there will be people willing to spend time in offering optimized MP3 sets in multiple languages. I for now like the script which lets me generate a whole set including map file in a couple of seconds
-stopwatch timer alerts: not realized yet...

One thing i noticed is i can no longer load main page config templates..When i do any testing i just use a empty model slot and set it up with one of the other page templates but for some reason it wont load...The load box high lights but it will not allow me to select a template...Not sure whats wrong

Do you have the same issue on current nightlies?