Placeholder Speech
Placeholder Speech gives every spoken line a stand-in voice, made from its words, so you can hear a scene, time it, and play it in your game before anyone records a line. Google’s Text-to-Speech makes it, on a key your team provides.
By default it sounds like a stand-in on purpose. It uses Google’s plainer Standard voices at a low sample rate, so nobody mistakes it for the finished thing. For a better read, a project can choose HD voices, which cost more. Either way, any real voice clip replaces it: a scratch voice clip, a recorded one, or a final one is always heard instead.
Turning it on
Section titled “Turning it on”Placeholder Speech needs Audio Folders. In Project Settings ▸ Placeholder Speech:
- Tick Generate Placeholder Speech. Patterpad adds a placeholder stage to your recording ladder, just
above missing, so its folder is
placeholder/under your audio root. (If your ladder already has a stage called placeholder, it uses that.) - Set the Google Cloud key file (see getting a key).
- Check the accent. It follows your project’s main language: a project written in
en-USspeaks with an American accent, and one written in plainenwith a British one, so choose another there if you want it. - Leave Voices on Standard, or choose HD for more natural voices that cost more (see HD voices).
Then carry on writing. A moment after you move off a line you’ve edited, its speech is made in the background, with a strip at the bottom of the window while it works. Lines still waiting when you close Patterpad are made next time you open the project.
To make a whole scene or the project at once, use Production ▸ Placeholder Speech ▸ Make for This Scene or Make for the Project. The project one says how many lines it will make first.
Voices
Section titled “Voices”Each character’s voice is in Project Settings ▸ Cast: open a character’s details to see Placeholder Speech voice, with Hear to listen to it. With no voice chosen, the character’s name picks one, so everyone has a voice from the start, and each character always gets the same one.
A character’s grammatical gender narrows the pick when it’s male or female: if the name picked a voice of the other kind, it moves to one that fits. Characters whose voice already fits keep it. Renaming a character keeps their voice.
Change a voice, a gender, or the accent and Patterpad offers to make that character’s lines again. If you’d rather wait, Production ▸ Placeholder Speech does it later.
HD voices
Section titled “HD voices”HD voices are Google’s Chirp 3 HD voices. They read a line far more naturally than the Standard ones, which helps when you play a scene to someone. They’re still placeholders: any real voice clip replaces them, and a line with only HD speech is still to record.
They cost more. Placeholder Speech is billed to your own Google account, and Google charges HD voices a higher rate per character than Standard ones, with fewer free characters each month. In November 2026, Standard voices were US$4 per million characters with 4 million free a month, and Chirp 3 HD voices US$30 per million with 1 million free. Prices change, so check Google’s Text-to-Speech pricing before you switch.
To switch, choose HD under Voices in Project Settings ▸ Placeholder Speech. HD has its own voices, so every character without a voice of their own gets a new one, and all their lines are made again, at the HD rate. The settings page says so before you save, and Patterpad asks before it starts making them.
To switch back, choose Standard. The lines are made again with Standard voices.
A character’s own voice stays as it is when you switch, even one from the other kind, and Cast marks it. An HD voice chosen for a character in a Standard project is billed at the HD rate.
What it means for the rest of the project
Section titled “What it means for the rest of the project”- A line with only placeholder speech is still to record. The recording script and Production Information count it as still to record, the scratch recorder’s Next needed stops at it, and the inspector offers ● Record on it as it does on a line with nothing.
- Editing a line makes its speech again, on its own. It isn’t flagged for re-record, since nobody needs to do anything. If Google sends a clip cut short (it happens now and then), Make again in the inspector asks for it again.
- Your game plays it like any other voice clip, through the audio manifest,
so a development build can be heard all the way through. For a release, leave the placeholder folder out of
what you package and write a shipping manifest without it:
patter export --audio-manifest-out build/patteraudio.json --no-placeholder-speech. - Its files are WAV, 16 kHz (24 kHz with HD voices), trimmed of silence at each end, as scratch voice clips are, so lines are timed by their speech and not by Google’s pauses.
- A file you put in the placeholder folder yourself is never touched. Placeholder Speech marks its own files, and only remakes or removes those. The settings page lists any it didn’t make.
Getting a key
Section titled “Getting a key”Placeholder Speech uses a service account key: a small .json file from Google Cloud that lets Patterpad ask
for speech on your project’s account. One person usually makes it, and the team shares the file.
- Sign in to the Google Cloud console and create a project for your game.
- Set up billing on it. Text-to-Speech has a free monthly allowance, and the Standard voices Placeholder Speech uses by default go a long way within it (HD voices less far), but Google needs a billing account first. Set a budget alert in Billing while you’re there, so you hear about it before anything is charged.
- In APIs & Services, enable the Cloud Text-to-Speech API.
- In IAM & Admin ▸ Service Accounts, create a service account. It needs no role.
- On the service account, open Keys ▸ Add key ▸ Create new key, choose JSON, and save the file.
Keep the key out of version control. A key committed to a repository can be copied by anyone who sees it (and
Google disables keys it finds in public ones). Put it in a folder outside your project, such as .patter in your
home folder, and point Patterpad at it with Choose…. Patterpad stores the path, never the key:
- A path under your home folder is stored as
~/…, which works on Windows, macOS, and Linux, so a team that keeps the key in the same place on every computer shares one setting. - A path relative to the project works too, for a team that keeps the key beside its checkouts.
- If the path is empty, or the file isn’t on a computer, Patterpad uses the standard
GOOGLE_APPLICATION_CREDENTIALSenvironment variable instead, and then a gcloud sign-in (below).
If you already use gcloud
Section titled “If you already use gcloud”Someone who works with Google Cloud from the command line can skip the service account and use their own sign-in:
gcloud auth application-default login, then gcloud auth application-default set-quota-project with the
project’s id, so Google knows which project to bill. Patterpad finds that sign-in where gcloud keeps it
(~/.config/gcloud/application_default_credentials.json, or %APPDATA%\gcloud\ on Windows) when no key file is
set, or you can choose the file itself. It belongs to one person, so a team shares a service account’s key instead.
The settings page says whether the key works on this computer, and which Google project and account it belongs to. If the key sits inside your project’s folder, it says so.
Without a key
Section titled “Without a key”A teammate whose computer has no key gets one quiet note per session, never an error. Their saves aren’t held up, and they hear the speech others made and committed. To stop the note on a computer that’s never meant to make speech, untick Production ▸ Placeholder Speech ▸ Make on This Computer.
A team can also keep the key on one build machine only, and make the speech there with
patter placeholder-speech.
Keeping it out of version control
Section titled “Keeping it out of version control”Speech is made from the script, so it can always be made again. To save repository space, tick Keep Placeholder Speech out of version control: the speech is then written straight to disk and never added, and its folder tells git to ignore it. Each computer with a key makes its own; a teammate without one hears none.
Open source under the MIT licence, made by Ian Thomas.