Voice alerts test build

Sometimes the solution can be so simple
Concerning more number files, it could be easily done, as a number mp3 only takes a couple of kB. Where would you want to start single digit readout? at 100? I'm open for user feedback on which solution is preferred, I opted for the simplest and most universal solution for starters

The deviation 5.0.0 firmwares are provided by vlad_vy, it is his personal choice to add any other commits.

Waiting for more feedback!

magic_marty wrote: Has the devo12 also been updated with the latest build?

Yes, all builds (devo 6, 7e-256, 8,10 and 12) have been updated and are on the same code base. 7e is available on request, it does not have the menu or telemetry value readouts due to Rom space issues, but it can parse music.map and music configuration in model inis fine, so you can manually edit the model.ini or use the music configuration menu in the devo 10 emulator.

Next on list is the addition of buttons and auctions knows.

What I was thinking about was :
As the programm reads the number, split them in different sounds that are assembled for one figure say nine + point + zero+ one for 9.01 to pronounce any figure it would be necessary to add ten, eleven twelve.........nineteen twenty, thirty, fourty.....ninety and may be hundred, thousand, to be able to pronounce any figure.

Some more testing on my devo 8
All the switches sounds work perfectly.
The low battery voltage works
The time remaining on a count down timer works. Sometimes some words are missing ex remaining time two (missing zero seconds) It would be good to have sixty etc instead of six zero etc. There is no time up. A suggestion. could it be possible to have more time for prealert and a possiblity of beginning for example at 2 or 3 minutes with a count down in minutes until one minute and by 10 or 20 seconds after. I realize that it is a change which may exceeds this topic but that would be nice.

Welcome message : I never heard the 0000 message but sometimes I heard the 2000.mp3 (4 or 5 times in a row and never after)
Binding message : never.
Saving message never
Good bye message : never
Inactivity alarm : never (and I would really appréciate this one cause I often forget to switch off my TX)
Some telemetry alarms are in the hardcoded zone (13 to 16) and some are in the custom ones What is the difference ?

All that is very good and promising !!!

For pre alert, I was thinking you can assign multiple alarms to the same (telemetry) event.

F.I.
First alert, battery getting low, <3.5V
Second alert, Emergency battery dead, <3.2V (or maybe just play a specific Sirene sound, continuously

But on alerts the upper- lower threshold (bandwith) and the time in ms that the criteria is valid be to be set.
also there may be some setting on preferences of alerts?
latest alert will be played > stop playing the previous alert ? Or play alerts after each other?

BTW, I have goodbye and welcome on my U7E, maybe has to do with the boot time etc of the TX, possibly some start-up delay to be added?

1.) startup message : this could well be possible. If possible hook up a logic analyzer to check if any data packets are set on startup to the ppm port. There is already some delay for the startup message, maybe we have to increase it.

2.) prealert for timers : the time before alert is set in the transmitter menu.

3) binding, battery and saving I haven't looked into yet.

4) telemetry alarms: the general alarms are used unless there are specific alarms defined for a model. Imagine setting up an telemetry alarm but forgetting to set a music, your alarm would go unnoticed if it wouldn't yplaythe generic alarm

5) I will try to fix and enhance the digits handling code for value readouts as suggested.

Unfortunately I have no logic analyzer.
As concerns prealert for timers the max possible is 1 minute which is good when a buzzer is used (no difference between 1 or 3 minutes) . But with words it would be nice to begin for example at 3 minutes. But 3 minutes by 20 seconds make too much alarms. So by minutes as long as there is more than one minute and by say 20 seconds after. I imagin it would be complex.

It can be for a glider effectively or a jet or a seaplane for example. Sometimes it takes long to prepare a landing in good conditions and i prefer not to be stressed.
It would be good also for races, even in 6 or 7 minutes races it is very good to have the remaining time each minute.

FDR wrote:

Moeder wrote: 5) I will try to fix and enhance the digits handling code for value readouts as suggested.

I have doubts if that is doable, or rather if it should be done. For example if somebody would like to play the sounds on his own languange, the articulation of a number will be different from english for sure, while reading out the digits ("three-point-one-four") will probably do in any language...

It works very well effectively in french (I am) "trois point un quatre" is very understandable for a frenchman. But twelve point is simpler than one two point. after effectively for other languages it may be different. But a lot of people speak english (especially in aeronautics )and why should not they use english if the articulation of their language does not fit with the scheeme ?

Got the test build installed on my Devo12 and all is working good as reported i do not get the "welcome and Goodbye" but i didn't get it before with the original MWM build unless i powered the DFplayer with a external power source..currently i am powering everything from the devo board...i noticed there is alot of mp3 files in the folder but i only get 6 available ...
How do we go about adding more sound files to be played via switches? For example on a flight controller if i want to add a voice file for "Return to Launch" or "Mode Stabilize" ?
This is looking very promising and thanks again for your work on furthering the voice development...
<Marty>

Hmm.. It wouldn't work in my own mother tongue either... I'd suggest sticking to the current scheme. How is this solved in opentx?

magic_marty wrote: i noticed there is alot of mp3 files in the folder but i only get 6 available ...
How do we go about adding more sound files to be played via switches?

The files 0000-1999 are currently reserved for preset alarms to not break any compatibility. I guess we could also move the number calls to the 0000-0999 range and start individual alerts at 1000,that sounds a bit more logical.

You can add your own entries to the music.map if you record your own mp3 and also get the file's duration. Or, if a Mac or Linux user you can use the genspeech tool in the utils folder of my repository to automatically create all desired files and the map file.

In the menu, only ids starting with 2000 will be displayed

Moeder wrote: Hmm.. It wouldn't work in my own mother tongue either... I'd suggest sticking to the current scheme. How is this solved in opentx?

In Open tx there are a lot of wav files... one for each figure from 0 to 100 and after 200 to 900 by hundred and one for thousand.
Effectively that makes a lot (110) but easy to do...and to translate by users if they want (I am speaking of the wav files, not the programming !)
translations exist in czech, deutsh, english, spanish,french, hungarian, italian, polish, portuguese, and slovak.

Prealert time: what if we just generally increase the possible pre alert time from 1 to 3 minutes in transmitter config? It really has nothing to do with voice alerts...

Numbers: let's stick with the universal solution for now. I'd rather put time and effort in realizing the rest, fixing bugs and cleaning up the code

Moeder wrote:

magic_marty wrote: i noticed there is alot of mp3 files in the folder but i only get 6 available ...
How do we go about adding more sound files to be played via switches?

The files 0000-1999 are currently reserved for preset alarms to not break any compatibility. I guess we could also move the number calls to the 0000-0999 range and start individual alerts at 1000,that sounds a bit more logical.

You can add your own entries to the music.map if you record your own mp3 and also get the file's duration. Or, if a Mac or Linux user you can use the genspeech tool in the utils folder of my repository to automatically create all desired files and the map file.

In the menu, only ids starting with 2000 will be displayed

That's what i actually want to do is be able to add custom entries to the music.map i just don't fully understand how to do it but plan on giving it a try and a little experimenting to figure it out...Perhaps this weekend..

Moeder wrote: Prealert time: what if we just generally increase the possible pre alert time from 1 to 3 minutes in transmitter config? It really has nothing to do with voice alerts...

Numbers: let's stick with the universal solution for now. I'd rather put time and effort in realizing the rest, fixing bugs and cleaning up the code

I agree with you on all that. As concerns prealert that has nothing to do with voice alerts. So another suggestion ( for the future ) : to have a voice alert just linked to the timer : an information given for example each minute spent on the timer.
And for numbers, The present solution is very good up to nine and there are other priorities. Just keep it in mind if it could be simpler. (I would agree to create the necessary mp3 files).

Regards

Creating the mp3 files with genspeech is just a matter of a couple of minutes. The problem is, that some languages say the digits in reversed orders, for example 21 in English is 20-1, in french it would be 20-&-1, and in German it is 1-&-20. So we would need to make quite different digits handling codes for different languages.