This program started as a fairly simplistic python program that watches the City of Heroes chat log file. When it sees a message from an NPC, it uses the windows text-to-speech capability to read out the message. It also notices when you receive a badge and tells you which badge you got.
It still does that, mostly. Instead of using the same built-in voice for everything, it uses different voices now. It supports a bunch of voice effects. Adding new effects is pretty easy.
When Sidekick starts you get a text window and a GUI. It takes a hot second to get going.
After you've started the game and are logged in with your character (and have followed the steps under 'Configure' to enable logging) click "Attach to Log" and you'll see something more like this:
Note the chat dialog echoing in the terminal window and the framed contextual help for "DFB".
As you play the graph will update and track your experience and influence gain over time. Its a good way to gauge your progress in the game.
You'll also start hearing things. This time the voices in your head are real. Whenever an NPC or a team member says something in chat, it will be converted into audio.
Don't like what someone sounds like? That is what the "Voices" tab is for:
Adding more voices is an easy way to increase the variety and flavor of sounds you hear in the city.
- Open Settings, choose Time & Language
- Choose Speech on the left side
- Choose "Add voices" under Manage voices
By itself that will probably only give you one or two more voices, even if you install a dozen. There is a powershell script in this git repo 'enable_all_win10_voices.ps' that copies voices from the "only use for windows transcribe" part of the registry out to the "use for anything" part.
I think you have to reboot or logout/login before the extra voices are available.
Multiple paid text-to-speech services are supported. For google you will need a valid application default login. https://cloud.google.com/sdk/gcloud/reference/auth/application-default/login
I know this is a pain. The quality and flexibility are much better than Windows TTS but it takes a moment to generate and the delay is annoying. Responses are cached to keep costs down to near zero.
ElevenLabs voices require an ElevenLabs API Token. You can configure it on the "configuration" page.
Very easy. The quality is really great but the 10,000 characters of text-to-speech you get each month with a free account can easily run out. How long varies dramatically with how much you play and which parts of the game. Somewhere between a 5-15 hours of game play. They have a $5/month plan than gives you 3x more.
Every voice can have audio effect layers applied to it.
For better or worse, the UI for manipulating this is immediate -- Whenever you change something it is automatically saved and if you are in-game you will start to hear the new voice/effects for any new speech. You can choose the bottom option in the phrases dropdown to:
"Rebuild all phrases". It will... rebuild all the phrases with the current voice configuration.
Players on your team get voices too. Find them in the list to modify what they sound like. When players talk it is NOT cached, that tickled by privacy bone and the value of the cache is minimal. Since these aren't cached I suggest just using Windows TTS voices for players. it's up to you, the settings are on the Configuration page.
Go to the releases page:
https://github.com/jason-kane/coh_npc_voices/releases
The release has one porky little pig zip. Uncompress it whereever you want it.
In the "sidekick" directory run "sidekick.exe" It should be pretty easy to find.
Go to the "configuration" tab.
You need to set the directory for the CoH log. It's important to get this right. This is where CoH stores the chat logs. Mine is:
C:/CoH/homecoming/accounts/VVonder/Logs
You also need to choose a directory where generated audio is cached (clip library). Thsi can be an empty folder, but you have to choose where to put it. If it gets too big you can just delete all or parts of it. They are just .wav files.
You see the tabs, "Google Cloud", "ElevenLabs", "Amazon Polly", "OpenAI", "Azure" ?
Each of those explains how to configure sidekick to use those services. You can configure all, none or anywhere in between. Totally up to you.
Below the instructions for each engine there is a grid for primary/second engine.
You choose which TTS service should be used for NPC, which for other players and which for system messages. This makes it easier to spread requests out over multiple providers so you can stay longer in the free tiers.
The sidekick will use whatever you choose as the "Primary Engine" until it wants to spend cash. Then it switches over to the "Secondary Engine".
Finally, you can choose how chatty you want the sidekick to be. I would start will all the boxes checked and turn some off if they are annoying.
Once you are configured, run the game launcher, login, pick a character, same as always.
After you are zoned into the game, click "Attach to Log" from the Character tab.
That is it. Enjoy the ride.
The sidekick doesn't always exit cleanly. I don't know why yet. Clicking on the terminal window then pressing control-c can help.
Go to the releases page:
https://github.com/jason-kane/coh_npc_voices/releases
Download sidekick_setup.exe and run it. Windows will probably complain since I'm not paying 300/year for an application certificate. The setup program is small (5MB-ish) but this thing downloads some heavy packages like numpy. Just a heads up in case your connection is slow or expensive. After the initial install it will be quick and upgrades to sidekick will not re-download unnecessarily.
This the first option on the configuration page.
It will vary based on the coh launcher/installer. Probably something like:
<COH Install Directory>/accounts/<Account Name>/logs/
<COH Install Directory>/homecoming/accounts/<Account Name>/Logs/
You will need to enable logging on your character inside the game.
You can use this slash command in the chat window:
/logchat
It's very possible you have to quit CoH and login again before the logging works. Now you are ready to "Attach to Log".
The "Character" tab currently shows a graph of how much XP/minute your character earned in the last hour of playtime. The samples are binned into one minute intervals and the graph updates once per minute.
Any NPC that talks will get a voice. If you don't like the way it sounds you can use the "Voices" tab, find the character that spoke and fiddle with it. You'll find a record of the things that NPC says in the dropdown so you can tune it based on their own words. It all saves itself as you go.
90% of this is totally compatible, no problem. The other 10% is the windows sapi stuff; making it platform detect and default to the (good) osx text-to-voice or the (not as good) linux espeak will take some effort but isn't particularly difficult. Its a longer-term TODO (PR welcome?). I don't game on Linux so I'm just not setup for it.
pip install git+https://github.com/jason-kane/coh_npc_voices.git
pip install --upgrade git+https://github.com/jason-kane/coh_npc_voices.git
pip uninstall coh_npc_voices
OBSOLETE
Start a cmd terminal and run:
sidekick.exe
Start City of Heroes; any launcher, any host. login and pick any character. Then enable chat logging.
You can use this slash command in the chat window:
/logchat
It is also in the settings UI somewhere.
This can be a per-character thing so if you aren't getting voices double check this first.
Next you will need to figure out where your logs are stored. It will vary based on the installer.
<COH Install Directory>/accounts/<Account Name>/logs/
<COH Install Directory>/homecoming/accounts/<Account Name>/Logs/
Use "Set Log Dir" in the upper right corner to configure it for your log location. You should only need to do this once. It is a little trixy because it needs to see the "Welcome to Paragon City" thing in the log. That means you probably have to enable logging, then logout/login, then "Attach to Log". It's annoying, but it only happens the first time.
Once you are ready You will need to click "Attach to Log" in the upper left corner after you are in-game. The "Attach" button changes to a "Detach". You can attach/detach whenever you want.
The right directory will have files with names like "chatlog 2024-03-31.txt". If you are having trouble turn on /logchat for a few minutes in-game, then look for file named after todays date.
If you are curious and want to poke around at the source code it is right here. You can download your own copy with "git".
git clone https://github.com/jason-kane/coh_npc_voices.git
cd coh_npc_voices
The python source is in src/coh_npc_voices. I've only made minimal efforts to make it nice. It is tied to tkinter more than I would like. The blend of multiprocess and threads is confusing; and core functionality is replicated for no good reason.
But it does work, and it is fun.
Obsolete
Getting inno to wrap a virtualenv in a portable way is hell. So I threw in the towel and switched to pyinstaller.
First run fresh.bat
fresh.bat
That will clean-slate the venv directory and apply some path detection tweaks. Next run innosetup, load this projects .iss file then -> Build -> Compile. Build -> Open Output folder will give you the dir with the sidekick_setup.exe.
The way this works is a little bit awesome. sidekick_setup.exe will install our files and a barebones python venv, then it will run win_install, which is a compiled version of win_install.ps1.
win_install.ps1 gets compiled into win_install.exe whenever 'fresh.bat' is run.
It installs (using pip) all our dependencies. End result? A small (5MB) setup executable that installs all the crap we need (I'm looking at you numpy. Try eating a salad). Running it again? No problem. Totally idempotent. If you use the same destination directory it won't even need to re-install anything that hasn't changed. The best part from my POV is the actual running code is sitting there for the user to poke at with no obfuscation.
I'm currently pleased as punch with the installer. Kind of hell to get it all figured out but the results are quite nice.
To restore the dev environment after cutting a release, run dirty.bat
Please leave an issue here in github. I can't promise anything but I'm very likely to read it.
MIT. I think that means you are allowed to do what you want, just don't blame me if it all goes wrong. That said I would appreciate pull requests if you make improvements so we can all share the benefits.
In many ways this is a UI wrapper around a mashup of these three modules:
https://github.com/austin-bowen/voicebox/ https://github.com/DeepHorizons/tts https://github.com/spotify/pedalboard
Big thanks and shoutout for the creators of these packages.
And these Text To Speech providers:
Amazon Polly
MS Azure TTS
ElevenLabs
Google Cloud TTS
OpenAI TTS
MS Windows TTS
Without cheap, high quality text to speech services this would sound like berzerk.
Raw entity data from https://cod.uberguy.net/html/index.html