Voice alerts test build

I floated the idea of an SPI-based USART because as you have noted we simply don't have available TX pins. The context of the four-in-one board that plugs into the CYRF port might, from a hardware perspective, simplify this feature and provide more universal acceptance. Perhaps even using an SPI GPIO board seems appealing to me. But there are drawback to this approach: interference with the high-speed data requirements of the protocols seems to be the most troubling.

Another idea I've been researching is a way to implement a "Soft-Serial" approach like I often use with my Arduino projects. I don't have enough programming experience to know if this is even viable, but it could expand our choices for pins. Is this something you could help look into?

Until we surmount these issues, or find new alternatives, we're left with using an unused TX pin or the trainer port.

@Deal57, I'm not familiar with soft serial, but I've heard of it on Cleanflight which is also using a similar STM32 MCU. Seems to me it's possible to borrow/implement it on Deviation. Maybe one of Gurus here can share some insight on the subject.

Added the following functions to voice alert:

• Support voice feedback on Aux Knobs & Trim button when use as virtual switches.
• Added new options "device" & "vibrate" in sound.ini file to allow selection of playback device and vibrator control on individual alert messages.

Bug fix:

• Missing welcome message when voice module is powered internally.
• Re-enable haptic feedback (vibrator).
• Misfired of Telemetry Alarm 1 on various nrf24l01 related protocols.

The new version has been pushed to the same repo under branch extended-audio. If anyone plans to test it, please report bugs if find.

phantom8 is there any way to get genspeach running on linux (or even windows :/)? or is it osx only?

I downloaded the extended audio branch file but not exactly sure how to go about installing it to my devo12....Do i just copy the files to the 12 in usb mode?

HappyHarry wrote: phantom8 is there any way to get genspeach running on linux (or even windows :/)? or is it osx only?

I've updated the genspeech utility to support linux. You'll need to install espeak & lame. The genspeech is a shell script and can be ported to other platforms. I'm not a Windows guy, but I think you can install cygwin to run linux on Windows. The new genspeech has been pushed to my repo. I've also attached a copy here.

magic_marty wrote: I downloaded the extended audio branch file but not exactly sure how to go about installing it to my devo12....Do i just copy the files to the 12 in usb mode?

I've compiled a Devo12 built with the latest nightly and attached here. A u7e built is also included. Both of these builds use PPM port for audio.

got it some what working but everything to do i get "telemetry alarm 6 triggered" lol except for powering on and off and the timers expiring...

magic_marty wrote: got it some what working but everything to do i get "telemetry alarm 6 triggered" lol except for powering on and off and the timers expiring...

What actions did you do to trigger the Telemetry Alarm 6? Try to test the Bind alert message by choosing a protocol that has the bind option.

Did you try to configure the switch voice feedback? I've attached mwm's voice files here. The volume on this version have been boosted.

not sure how to configure the switch feedback the model bind does announce when it binds...

magic_marty wrote: not sure how to configure the switch feedback the model bind does announce when it binds...

So, I suppose the system alerts work, but I'm not sure what exactly triggers the Telemetry Alarm 6. It seems to do with the protocol that you're using.

Do you have access to a OSX or Linux machine? If you do, you can download the genspeech utility in a few post above. There is a readme.txt file in there and it will tell you how to generate speech files for you models.

unfortunately i don't have any OSX and very very minimal experience with linux a few years ago...Perhaps i will search to see if i can find a windows speech generator...the timers countdown works as well i set up a test model file and used the mixer set to the trainer switch and the mux to beep and it says telemetry alarm 6 triggered when toggled..just not sure how to assign sounds to actions ect..

magic_marty wrote: unfortunately i don't have any OSX and very very minimal experience with linux a few years ago...Perhaps i will search to see if i can find a windows speech generator...the timers countdown works as well i set up a test model file and used the mixer set to the trainer switch and the mux to beep and it says telemetry alarm 6 triggered when toggled..just not sure how to assign sounds to actions ect..

The alerts that you heard so far are the system pre-programmed alerts. I'm not familiar with telemetry alarm, I only know they're protocol related and will trigger when certain criteria reached.

If you want to test out the switch voice feedback, you can prepare the model alert message files (check the readme.txt on how to prepare them) and send them to me via PM. I can generate the speech files for you to test.

There are also some web based text to voice tools if you want to generate your own and don't run something that at least reminds you (or me ) of UNIX. I provided a pointer to the one I used in the original build.

The telemetry alarms are set on the telemetry page. There are six of them, and the first one on the page triggers alarm #1, etc. But you have to enable them - they aren't on by default, and can't be turned on if telemetry is disabled or nor available for that protocol. The alarms themselves don't depend on the protocol, but the set of available measurements that can trigger an alarm do.

Is it possible to use the openTX voice alerts with this ? They're pretty good and it might save us a bit of work

Nitro_123 wrote: Is it possible to use the openTX voice alerts with this ? They're pretty good and it might save us a bit of work

You can of course use the OpenTX voice files, as long as they're in mp3 format. If you want, you can even use your own voice recording. Just upload the mp3 files to DFPlayer mini and rename them properly, such as 0100.mp3, 0101.mp3, etc. Then, you need to provide model mapping files to define what switch maps to what mp3 files. For details, refer to the readme.txt.

phantom8 wrote: I've updated the genspeech utility to support linux. You'll need to install espeak & lame. The genspeech is a shell script and can be ported to other platforms. I'm not a Windows guy, but I think you can install cygwin to run linux on Windows. The new genspeech has been pushed to my repo. I've also attached a copy here.

many thanks, linux is perfect for me

Seems the problem i am having is figuring out how to do the "model mapping files to define what switch maps to what mp3 file" i just don't see them for some reason in the files of file system ect...

if you download the source from github and look inside the osx/models folder there are 2 pre made alert files to look at, and if you read the osx/readme.txt file it has all the info you need to get it working by writing your own model alert files, you then run genspeech and it will make the mp3 and map files for you using the alert files as the basis.


phantom-8 I'm still waiting for the dfplayer to arrive but I've had a play around with genspeach using espeak on linux and all is working well in regards to producing the mp3 and map fikes, espeak is pretty good also with a littke bit of tweaking, especially using the en-sc voice lol. thanks to you and mwm for sharing the code

yes i downloaded the source from github and read the readme files but i can't run genspeech on windows..

not sure if there is a windows program that will make the map files

@HappyHarry, thanks for taking the time to help answering questions. I haven't played much with espeak. The default voice is pretty robotic. I'll try the en-sc voice that you suggested.

@magic_marty, if you like, you can send me the model alert message files and I'll generate the voice files for you. If you can make the voice files yourself (such as using the online text to speech utility that mwm had mentioned), you can prepare the switch mapping files yourself. The readme.txt file, point 3 descripts the format of the map file. Basically, you name all your voice files as 0100.mp3, 0101.mp3, 0102.mp3, etc. Copy the mp3 files to your DFPlayer Mini SDCard's mp3 folder. For each model that you want to have switch voice feedback, prepare a mapping file, such as "model1.map".

File: model1.map
Content:
In the mapping file, you type the name of the switch, followed by a colon. Then, the voice file number that you had prepared for the DFPlayer mini. After you're done with these mapping files, copy them to your Transmitter's models directory. Hope this help!