ERSky Voice Files

[Mutchy]

I want to set up voice alerts and voice repetition of telemetry values.
I have read Mikes description of how to set up the voice alerts in the October 1st post above. I have downloaded jhsa's voice files from September 12 above and installed them in a folder called "voice" in the root directory of an SD card and installed it in my ErSky9x board. The menu says that the SD card is "ready 0008" followed by a bunch of hex addresses. When I go to the Safety Switches page I get the options of "S" and "A" but no "V" option in the first column. Nor is there more than 16 Safety Switches. My version of ErSky9x is r72 ver 706-Mike of 27-05-2012.
Do I need a later version of ErSky9x? I can only find a rev 80 which is later than mine and it is a test version. Is the latest here?
http://ersky9x.googlecode.com/svn/trunk/ersky9x_rom.bin
The thread Mike links to above is 70 odd pages long. Is there a concise explanation of how the voice files work and how to implement them? For example I gather that they can be concatenated so that one .wav file is spoken after another but how is that done on the Safety Switch page or on the Custom Switches page?
Steve

[kaos]

the latest version of ERSKY9X is here: r122 v1375

here: http://code.google.com/p/ersky9x/

[Mutchy]

Thanks heaps kaos.
What's this page for then?? why doesn't it have the latest version?
http://code.google.com/p/ersky9x/downloads/list
Steve

[kaos]

ERSKY9X is still under development. A lot of changes in a very short period of time. Unfortunately, Mike is really short of hands. Just about doing everything by himself alone. The wiki page you listed is what considered 'stable version', not quite up to date. Also, the r122 is sort of 'test version'. I tested it without problems besides those things still can be developed/improved.

by the way, here is a copy my ersky9x voice file on my SD in Tx. I think it is as complete as I can get. May help you save some time. you want to copy all these wav files under a folder 'VOICE' in the root directory of your SD card.

program the wav file '0269' and hear it.

[Mutchy]

Thanks kaos
I have loaded r122 successfully. I know because all my models are voice annunciated as "Easy Star"!!!! (ie .wav file #260 from jhsa'a voice files) I'll find out more later tonight when I have a chance to look at the menus. Then I'll have a listen to .wav file 0269!

Some notes for beginners like me:
The ersky9x_rom.bin file needs to be in the same folder on your hard drive as the SAM-BA program. Go to "my computer" and double click on the "C" drive. double click on the 'Program Files' folder. Then double click on the 'ATMEL corporation' folder. Double click on the samba_2.11 folder. There you will find the samba run time executable (the icon has drawing of a chip on it) together with the tc-lib folder (where you put the usb driver software when installing the samba usb driver) and the uninstall program. drag and drop the ersky9x_rom.bin file into this folder then follow the instructions for starting SAM-BA here:
viewtopic.php?f=7&t=545
at the second post by SkyNorth.
When you have loaded the .wav voice files into the root directory of the SD card correctly, the first screen, 'long left' on the Tx menu, has an option for setting the sound mode. Set this as "speaker voice". In the second screen "long right", on the second line is the place where the .wav file for the name of the model is defined. The line "Voice Index" is set to the number of the .wav file corresponding to the name of the model. In the folder with the .wav files there is a text file which describes what sound is included in each .wav file.
Thanks again kaos.
Steve

[Kilrah]

Some notes for beginners like me:
The ersky9x_rom.bin file needs to be in the same folder on your hard drive as the SAM-BA program.

Or you can just use companion9x, and it will call SAM-BA to use your file wherever it is!

[kaos]

wav 0260-0280 are for model names. you can make your own model names. The ones in my zip files are what I use. You will change those per your need.
you can see the doc file in my zip that is the explanaiton of each wav file. pay attention to those color coding. red are system wav files, not to be changed. All others, you can modify as you wish.

each model set up there is a number right below the model name. that is where you set up your model name voice file. I think it defaults to 0260. that is why all your model announces the same name before you change it.

[kaos]

Some notes for beginners like me:
The ersky9x_rom.bin file needs to be in the same folder on your hard drive as the SAM-BA program.

Or you can just use companion9x, and it will call SAM-BA to use your file wherever it is!

I think you can specify where the FW is in Sam-ba before you burn it too.
But, use C9x to burn the FW is a lot easier. However, at the present time, I think C9X is not able to backup the ERSKY9X r122 eeprom correctly (mainly the variable voice switches) yet. Last time I tried , still need to do it manually using Sam-Ba.
But Romolo may have updated that in the last wk. I am not sure. I have not tried the latest C9X released in the past 2 days. You can try it to see.

[Mutchy]

Thanks for the explanation of the colours of the description in your .doc file for the .wav files. I have played your 0269 file and it's way cool! Er9x rocks!

I have used the Safety Switches page on my Tx menu to play many of the sounds and am impressed. She has a nice voice. But how do I stop the repetition? I can see where to define the 8, 16 and 24 seconds interval but not the number of repetitions. Is that done using a custom switch?
Also I set the TIMr to speak the elapsed count - up time. It works well for minutes up to four where I stopped the test but for any number of seconds greater than 19 up to 59 it did not announce the tens of seconds or the digits. It said for example "two minutes __ seconds" when the elapsed time was greater than 2;19 secs.
Now to find out how to make a .wav file.
Thanks
Steve

[MikeB]

If you set a safety switch to 'V' then you get 'alarms' voiced. These repeat as long as the alarm condition is present (switch is ON). The idea here is that you define a custom switch as, say:
V<ofs BATT 8.9v
This switch will be ON if the Tx battery is below 8.9 volts.
Now if this custom switch is used in a safety switch set to 'V' you get an alarm voice message that repeats.

Alternative use of safety switches.
At the top of the safety switch menu is an option to set a number of 'voice switches'. Suppose you set this to 3. The LAST 3 safety switches will now become VOICE SWITCHES.
Now you select a switch and a mode. The mode may be:
ON
OFF
BOTH
15 secs
30 secs
60 secs
Varible

ON plays the voice file once when the switch turns on.
OFF plays the voice file once when the switch turns off.
BOTH plays the voice file once when the switch changes.
the 15, 30 and 60 second options play the voice file at the selected interval as long as the switch is ON.

Varible lets you select a telemetry or radio value (battery, timer), that is spoken once when the switch turns ON.
If you want one (or more) of these repeated at intervals, then you can set a custom switch in timer mode where it is ON for a time and OFF for a time. Then you can use this custom switch as the switch selected in a voice switch to control the varible(s) being voiced.
The custom switches also have an extra AND switch now, so you can set a custom switch to to timer mode, with an override switch. Now the custom switch is only on while the override switch is ON and the timing sets it on. The AND switch also holds any custom switch set in timer mode in the OFF state, so it turns on within 1 second when the AND switch turns ON.
This all lets you have telemetry voiced at intervals when you have a switch allowing it.

I'll check out the time announcements, we did change which voice files were used and I probably haven't updated this hard coded option.

Mike.

[kaos]

Mutchy:
here is a detailed guide to voice switch porgraming written by Jhsa. It is embedded under hardware mod /audio mod. It is written for er9x but so far it also applies the same to ersky9x.

http://openrcforums.com/wiki/index.php/ ... io_on_er9x

By the way, just to let you know, the wav files I posted is mainly based on Jhsa's work then mixed with a few other's wav file ie playmodil and some of my own where I think the wav file sounds most pleasant to me.

Here is a link of ATT Natural Voice, you can create your own wav file easy, choosing different 'speaker' voice as you prefer. this is recommended to me by ReSt. Really easy to use. All those model name is done through it.
I found out you may want to add some spaces, commas, periods in the word or between words to make them sounds better. you can 'preview' it before downloading.
keep the number of wav file you try to make under 15-20 each time. Other wise it will stop working and asking you to purchase the program. Then just stop, quit the page, then relog on to the page to continue (probably clear the cookies 1st).
The wav files made through it is 16KHz, 16 bit it won't work in ER9X for modded stock 9x board but it works perfect in ERSKY9X for the SKY9X board.

http://www2.research.att.com/~ttsweb/tts/demo.php

[Mutchy]

Thanks Mike and kaos. I will use your valuable information after I write this short message.
I found that the original voice files from jhsa did not have numbers from 20 to 99 so i loaded the .wav files from kaos voice file and the time count-up now annunciates all the seconds (at 8/12/16 second intervals up to 59 of course)
I made my own model names and some other announcements using my own voice and voices and features which came standard on my Mac and then converted the .m4a files to .wav using Switch Sound File Converter. "Switch" appears to be slightly unsatisfactory. While the bit rate can be set to 8 bits the frequency defaults to 'auto' even if you select 16000 in the drop down menu. The names are reasonably good but there is a hollow sound - a reverberation - in the pronunciation of "seconds" in the timer count-up. I will write a full description of how to make model names and other announcements on a Mac later for everyone's information.
Steve

[Mutchy]

Ok. I've got the hang of defining the number of voice switches in the Safety Switches menu and have the ID0/1/2 switch announcing race mode/thermal mode/landing mode and the gear switch announcing screen ON/Off. Thanks Mike. I'll have a go at telemetry in a minute.
I read kaos post again and note that 16 bit works on ErSky9x. I made all mine 8 bit but I don't know what frequency they are but they sound OK. Your words sound a bit louder than mine, though, but then the voice is is slightly different. I'll convert them all to 16 bit and see what difference that makes.
Thanks for the link to ATT. I have a feeling that "Switch" will soon stop working unless I pay up and I am not happy to rely on "auto" for frequency and channel selection.
Steve
Edit: ATT is for Windows and Linux. Mac users don't need it. I'll find another audio file converter when "Switch" quits.

[MikeB]

I use Audacity, I think Jhsa does as well. It appears to be availale for MAC. With Audacity you can amplify a voice file so it has the maximum amplitude, and make all files the same volume.

Mike.

[jhsa]

I found that the original voice files from jhsa did not have numbers from 20 to 99 so i loaded the .wav files from kaos voice file and the time count-up now annunciates all the seconds (at 8/12/16 second intervals up to 59 of course)

Hey Steve, which files of mine don't have the numbers from 20 to 99?

I've just downloaded them all and I can find the numbers there

Also my new files must work on the ersky board..
please check here and scroll to the end of the first post

viewtopic.php?f=64&t=1290&p=18110#p18110

Joao

[Mutchy]

Hello all.
I used a file from an earlier jhsa post in this thread and those .wav files are the bulk of what I still have installed. I did not know of the file at the link you posted above. It's very difficult to find that sort of information and to know whether or not it applies to your board when you start late as I have done. Of course they play beautifully. I just added kaos's .wav files from 0400 to 0499. I have no idea what went wrong and it was probably something I did. No criticism intended.
I'll have a look at Audacity and see how it compares with the way I made my voice files. (see next post)
Steve

[Mutchy]

Making a voice file on a Mac.

You can make a voice file using your own voice or one of the Macintosh voices using apps you already have that come for free with Mac. You then have to convert it to .wav format using an app like Switch File Converter:
http://download.cnet.com/Switch-Audio-C ... 03967.html

Using your own voice:
Open Quicktime Player. In the drop down menu 'File' select 'New Audio Recording' A graphic similar to an audio playback graphic will appear. It has a red dot in the middle. Click on that dot and speak to the installed microphone at the top of the screen. Click on the dot which is now a black square when finished. Save the file to the desktop. It is in .mov format. Open Switch Audio Converter. Drag and drop the .mov file into the big space in the middle of the application. At the bottom left corner select .wav as the output format. At the bottom of the page is a button labelled Encoder options. Select 'Custom' as the 'Settings' option. Then select 8 bit for Er9x use or 16 bit for ErSky 9x use, and click OK. It doesn't matter what else you select they will revert to 'auto' anyway. No I don't know why. Select the location where you want the file to end up and then click on the 'Convert' button. That's your .wav file finished. It will play in Quicktime.

Using a Mac voice.
Mac have had text to speech capability since about 1988. Highlight any text or list of names and numbers and press a keyboard shortcut ( 'apple' + 'esc' in my case) and it will speak the text. (your mac may have a different keyboard shortcut - see later)

In short to record a .wav file for your SD card: open iTunes, open TextEdit,. Open Switch. Write your annunciation (ie model names) on a single line in TextEdit, highlight it and press 'ctrl' + click the mouse (or "Right Click" on your trackpad), from the menu select 'Add to iTunes as a spoken track'. Go to iTunes page 'Recently Added'. As detailed above, drag and drop the .m4a file into the Switch window. Select the .wav format. select the custom encoder option and 16 or 8 bit . Convert the file and you are done.

Now there are many Mac voices and the default (Alex) is probably not what you want as it does not match what others have already recorded because they are all female and this voice is male. Go to System Preferences and open the Speech Preferences. Depending on the age of your Mac there will be a variable number of voices and the quality will be variable. My ancient MacBook has 5 male,5 female and ten novelty voices. The male and female voices have a 'Stephen Hawking' monotonic sound and could be used, but don't match what has already been done. My iMac with OS 10.8 has only 6 voices total of much better quality, but you can download up to 20 or so specifically for different languages ie an Italian voice for italian language text to speech. I used the Australian English voice 'Karen' which is a slightly softer but close match to what has already been done and has no Stephen Hawking effect. In recent OS's you have to download your voice from a list under 'custom' in the list of voices.

Your shortcut to speaking text and adding text to iTunes as a spoken track may be different depending on the age of your Mac. Look in System Preferences - Keyboard Shortcuts - Services . Here you can tick the box for 'Add to iTunes as a Spoken Track' and define your own shortcut. This command line will then appear in your ctrl + click menu.

Here's a link to help explain system settings for various operating systems:
http://osxdaily.com/2011/06/29/convert- ... -easy-way/

Steve

Edit. Here's a link to more information on the system voices available in various operating systems:
http://www.macworld.com/article/1164787 ... voice.html

To change the default system voice on older Mac OS's you may have to go to Applications > Utilities > VoiceOver Utility > Speech. This is where you set the default system voice.

[Mutchy]

The 16 bit files do sound less slurred than the 8 bit but not so different that I would go through and change them all.
Steve

[jhsa]

I'm making a sound pack using an online voice to speech thingy and splitting the files in audacity..
I think it will come out quite good.. will post about it here and on the audio thread. It might help someone..

[Mutchy]

I had a play with Audacity for the Mac. It is much more powerful than Switch which is just an audio converter; a Ferrari compared to a Fiat. It is as easy to use as Switch ie just drag and drop, then export as 16kHz PCM ,wav file. It is easier to use than Garage Band which is a Bugatti Veyron in comparison: a full musicians audio desk. It has an amazing number of options for modifying tracks, eliminating noise adding echo etc. which I will have to learn about. It is a very useful application which I will probably use instead of Switch because I can adjust the volume of the track and clean any unwanted noise.
And it is allegedly free!
Steve

[jhsa]

It is free.. I use it as a musician for my work..

You may want to look at this feature.

analize>sound finder
It will split a wav in multiple files..
then when you're happy with the split (you can adjust the label track), select "export multiple" from file drop down menu..

I used this to generat my latest sound files

http://www.acapela-group.com/text-to-sp ... -demo.html


Joao

[Mutchy]

Cool!
I like Audacity very much now that I have played with it a bit more; lucky there is a 'help' file!

I made a demo track for you, Joao. I hope you enjoy it!

0260.wav

(176.1 KiB) Downloaded 5529 times

Steve

[jhsa]

lol it sound like windows

[Mutchy]

I have an unwelcome sound generated by the hardware, I believe. It occurs at the end of the spoken words. Its a Bzzzt sound. Does anybody else have this sound?

P1010001.mp4

(1.07 MiB) Downloaded 5504 times

[kaos]

hmm, at one time there was a 'buzz' issue. But Mike/Bert have fixed it. Since I have no buzz.

Do you have the same 'buzz' playing jhsa or my wav file?
which program/method you use to generate your wav file?
Look into the wav file in audacity to see if there is a long silence wav at the end. get rid of that silence portion may help.

Nice looking Tx box. how /what did you use to make it?

[Kilrah]

Last I remember of that issue it was a problem with the files/format of the card.

[Mutchy]

The buzz happened only with my files, and only when operated with a switch. These were files which I generated using a Mac voice and Audiacity software. I had a period of silence at the end of the tracks. Deleting that silence has fixed the problem.

I built the Tx myself using parts from a JR 388 Tx and of course an ErSky 9x board and screen. Details here:
http://www.rcgroups.com/forums/member.php?u=24329
Click on the 'continue reading...' at the bottom of the page.

[kaos]

Alright! No more buzz.

that is a heck of a job to make that tray. I doubt I have the patience or the tools to make one. Nice job.

[Mutchy]

Here's my sound files for those of you who may have a Mac and could use a head start in creating a list of sound files for your transmitter. As I have said above they use the voice of Karen which does not have quite the clarity of the voice that kaos and jhsa have used but is more pleasant to my ear. Of course I am biased. She is allegedly an Aussie. I have eliminated all unnecessary quiet periods both before and after the sound so as to avoid the annoying bssst and click noises. This also seems to get rid of the "hollow drum" effect where it sounds as if the voice is inside an echoic chamber.
If you would like some help to make your own model names - with or without the introductory Macintosh or Windows start - up sound - just ask here.

I am extremely grateful to Mike B, kaos and jhsa for their tutorship, time and patience. Thank you very much.

The following advice from jhsa was very welcome:

"You may want to look at this feature.
analize>sound finder
It will split a wav in multiple files..
then when you're happy with the split (you can adjust the label track), select "export multiple" from file drop down menu."

Use of that feature in Audacity saves hours of time, allows automatic cropping of the quiet periods both before and after the sound and multiple exports in the correct sequence. Very very useful! You type up all the sounds you want to record in one long list,highlight the lot, control+click and select "add as voice file in iTunes", drag and drop this voice file into the Audacity new file window, set the project rate at 16000 Hz, select "analyse sound" from the "Analyze" drop down menu, set the before and after quiet periods to 0 seconds, and then "export multiple".

Mutchy

[Mutchy]

In the voice files there is an announcement which says "inactive". My Tx started saying that word today. I don't know what initiates that warning nor do I know what I did which eventually turned the voice off.
Could someone please explain.
Mutchy