Offline voices with Piper
The local voice is the one that works on a train, on a locked-down laptop and with nothing leaving your PC at all.
Speakybara can speak through three services: Piper on your own PC, Microsoft Edge's neural voices, and Google Cloud Text-to-Speech. Piper is the only one that is fully offline, and it is the default answer to "does anything leave my computer": with Piper, nothing does.
The image never leaves your PC in any mode, because the text recognition is the one built into Windows. What differs per service is the sentence text: Edge and Google voices send the sentence to Microsoft or Google to be spoken. Piper does not send anything.
Setting up a local voice
Step 1
Choose Lokaal (Piper) as the service
Open Speakybara, go to Models, switch on Your own models and keys and pick Local (Piper) as the voice service. Five voices are installed from the start: three Dutch and two English.
The card of the chosen service: a voice per language, and the button to download more. Screens come from the app itself.
Step 2
Download more voices if you want
The Download button in the Local (Piper) card offers about 10 more. Each voice file is around 60 MB and downloads once.
Step 3
Pick a voice per language
Dutch and English have their own dropdown, because Speakybara detects the language of your selection automatically: per selection, per sentence and per word. The play button next to each one speaks a sample.
Step 4
Set when the voice is unloaded
A Piper voice is loaded into memory while it speaks and unloaded after 90 seconds of silence. You can change that delay in Settings under Speech, Unload local voice after: longer means the next sentence starts faster, shorter means less memory in use.
Speech: speed, volume, and the delay before a local voice leaves memory.
Worth knowing
- None of the four voice services uses your graphics card: 0 MB of video memory, deliberately.
- With Edge or Google selected, nothing is loaded locally and the app stays around 55 MB. With Piper, add the voice file while it is in memory.
- If an online service drops out mid-text, Speakybara carries on with the local voice and tells you once, not once per sentence. When the service comes back, it switches back.
- Piper sounds noticeably more synthetic than the Edge or Google voices. That is the trade for being offline and free.
If something goes wrong
With Piper, yes, completely: recognising the text and speaking it both happen on your PC. The Edge and Google voices need a connection, because the sentence text is sent to Microsoft or Google to be spoken.
That is the voice file being loaded into memory, about 60 MB. It stays there until 90 seconds of silence have passed; raise that delay under Gedrag if you read in short bursts.
You do not have to choose. Speakybara detects Dutch and English per selection, per sentence and per word, and uses the voice you picked for that language. For English words inside a Dutch sentence, see the guide on your own English words.