Advanced Audio Suite for RC Transmitters

This Powershell V7 script allows you to use Microsoft Azure Speech Synthesis (Text-to-Speech) to generate custom, high quality Voice Packs for RC Transmitters.

Sample audio packs for those unable or unwilling to use the script are available here. They are all formatted for FrSky's Ethos 1.5+ operating system, but will likely work without issues on other systems, possibly with some modifications to the folder structure. The sample audio is based on FrSky's en.csv from their own audio generation system, but with my own added files that I feel are missing from their list.

All sample audio packs are generated with the following settings unless otherwise noted:

1. Style: None
2. Speed Multiplier: 1.25x
3. Leading Silence: 0ms
4. Trailing Silence: 25ms

Prerequesites

1. A subscription to Microsoft Azure (Free F0 tier will suffice)
   
2. PowerShell V7 (will not work on V6)
3. All prerequesites for Speech CLI Quickstart "Download and Install" Section
   
4. Speech Service Resource Group. See Speech CLI Quickstart "Create a resource configuration" section. You can skip the "To configure your resource key and region identifier, run the following commands:" section because this script does those for you (see Usage below)

Notes

1. This script is written and designed to work with any language supported by Azure TTS. Languages supported by Azure but not by Ethos may not work without some changes to the folder structure (untested).
2. Only one .csv file may be in the 'in' folder.
3. Thanks to Bender for his help in getting this to work.

CSV File

The script is designed for and needs three features in your CSV:

Filename:

Should be a two- or three-letter language code, such as en, fr, de, etc. e.g.: en.csv

Contents:

Column 1: path to file to be generated. e.g. system/throttle-hold.wav, system/fm-1.wav, gearup.wav, etc
Column 2: text to convert into speech. e.g. Throttle Hold, Flight Mode 1, Gear Up, etc

Downloading

Download the entire repository as a ZIP file. Extract the contents to anywhere of your choosing.

Configuring

You need to set your key and region for Azure before the synthesis will work. To do this, open the .ps1 file with an editor of your choosing (Notepad++ or VSCode are my recommendations).

At the top of the file, find this section, and replace yourkey and yourregion with your own information. Get this info from your Azure Portal TTS Resource Group.

spx --% config @key --set yourkey
spx --% config @region --set yourregion

Usage

1. Ensure your have your .csv file located in the 'in' folder. The downloaded ZIP from this repository includes my customized en.csv based on the Ethos 1.5 audio pack, but you can change it or update it as needed.

2. You should do the following to make it easier to run. The script is a .ps1 script. By default, if you just open it, it will open in a text editor (or ask you how to open it if you don't have a suitable editor). To make it easier to run in the future, right click the .ps1 file, click Open With, Browse for an App on this PC, navigate to C:\Program Files\PowerShell\7, and choose pwsh.exe. Click Always so that it always uses this method to run.

3. On the first run, it will create a key and region file using the info you put into the top of the script. Do not delete them, they are necessary for TTS to work and will simply be re-added the next time you run the script.

4. Once the key and region files are written, it will begin reading your .csv and the voices.json file. The voices.json file is a list of voices supported by Azure TTS and may occasionally be updated on this repo.

5. Once it's gathered the info, it'll look for a config.json file. If this is the first time running the script, there won't be one, so then it'll ask you to enter your settings for the first batch synthesis, starting with the compatible voices that match the language code from your .csv, e.g. en), followed by style (if applicable), speed multiplier, leading and trailing time, and Ethos version.

6. Follow the on-screen prompts. It's fairly straight-forward. Once all config options are entered, it will save your settings to a config file so that you don't have to enter them every time.

Common Errors:

Error 429: The Azure F0 (Free) pricing tier is limited to 20 requests per minute. This script runs considerably faster than that, so every once in a while you will get Error 429. Just let the script keep running and it will retry in a few seconds, up to 10 times before skipping.

Error 4429: You have exceeded the number of characters per month allowed on your subscription tier (Free is 500,000/mo). You will not be able to generate any more audio unless you change your subscription type or wait until the next monthly cycle.

Output

Once the script is completed, the synthesized audio can be found in the 'out' folder, according to the language code (e.g. en-AU, en-US) that you used for synthesis.

Other Notes

1. The script detects changes in your .csv file compared to last run. If it detects either a changed text to play on an existing file, or detects a new row, it only runs sythesis on the changed/added rows.

2. If no changes in your csv are detected compared to last run, it will re-run all rows but only files that have been deleted or were never synthesized will be synthesized.

Future Goals

1. I'd like to make it more of a menu system rather than just typing in stuff
2. I'd like to add support for the "Options" column (column 3) in the .csv. Not sure what it'd be used for, I guess that remains to be seen.
3. I'd like to support MultiLingual voices (such as AndrewMultilingualNeural) to allow generation of different languages with the same 'voice'. Currently, if you choose a language other than English, the en-US-*MultilingualNeural options will not show up, even though they can speak that language.

Any suggestions/feedback are always welcome.

Changelog

0.1 - Initial Release.
0.5 - Huge update

1. Added detection of changed or added rows in csv compared to the last run
2. Vastly improved configuration routine
3. Improved and shortened main loop
4. added selection between Ethos 1.4 or Ethos 1.5 format (or neither) and output folder is set according to this config setting
5. Cleaned up code to the best of my (very limited) ability