Skip to main content

Leer en español · Lire en français · Auf Deutsch lesen

Grok mark and a microphone, with Assistant underneath

Grok Assistant

Release candidate 1.0.5.

This is a voice companion for the moments when reading is difficult and the house is quiet. The laptop version is for an older person who already has a small computer nearby. The Raspberry Pi 4 version, with 4 GB of RAM, is the same assistant as a dedicated device, for a family that would rather not leave a laptop open. That parallel project is Grok Pi Assistance. The microphone can stay ready. The audio stays on the computer. Only text that was meant for the assistant is sent out.

I spent years waiting for the Amazon Echo to become a better listener. It never really became more than a speaker with a light ring, so I decided to build my own.

In my case, that person is my father. His eyesight is limited, and he spends many hours on his own. I wanted him to have a voice he could simply talk to — one that could answer questions and explain things without making him find a screen or read small text. That is now possible. Grok on this machine is also the foundation for the features and integrations I plan to add next.

For the assistant to hear properly, record a voice print on the microphone you will use: People → Prints. Without that print the program hears the room and drops the phrase, before the local model and before any order. The print is there so two people in the same place, the television, and the assistant's own voice do not mix into one conversation. The microphone stays open while the assistant speaks. Another microphone needs its own recording. The rest is in Voice prints and listeners.

While the program is running, it can keep listening. Speech is turned into text on this computer. Grok only receives text when the assistant has actually been addressed: after a greeting, for a question, for a command beginning with comando, or when someone asks for a song. Ordinary conversation stays in the local session. If a command is unclear or misheard, Grok can help interpret it, but the assistant still asks for a sí before doing anything.

That is the basic idea. Below are the listening window and the icon menu, with Escucha open. The pictures are the same in every language.

The window waiting, on the Simple tab

The icon menu, with Escucha open

The diagram below shows what happens to a spoken phrase from the moment it is heard.

The STT engine turns speech into text. Before a conversation, the local model checks for a complete command. During a conversation, it notes the line and the text goes to Grok.

In Settings, Grok only searches the internet. An administrator can allow Grok to change files on this machine. The shell stays off, and a relative path lands in the assistant data folder.

Read on

Metadata

Release files for grok-assistant 1.0.5

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for grok-assistant 1.0.5
File Size Uploaded
grok_assistant-1.0.5.tar.gz 270.0 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for grok-assistant 1.0.5
File Interpreter ABI Platform
grok_assistant-1.0.5-py3-none-any.whl Python 3 none any Details

Total release size: 541.2 kB

Release files / grok_assistant-1.0.5.tar.gz

Download URL grok_assistant-1.0.5.tar.gz
Size 270.0 kB
Tags Source
SHA-256 checksum
How to use checksums
647e3340481162fa40ab7a8d234231b2c7f9beec29350d88bce2636ecc47b8fa
BLAKE2b-256 checksum
How to use checksums
6f57d8a304c49553bf91bc197131805ed0a7cc39c952842f91c6bf950a3af32a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.14

Release files / grok_assistant-1.0.5-py3-none-any.whl

Download URL grok_assistant-1.0.5-py3-none-any.whl
Size 271.1 kB
Tags Python 3
SHA-256 checksum
How to use checksums
1b8f43f3112dc8ff7395b239e072829885a852e88f9239a9b644ffbaeee89508
BLAKE2b-256 checksum
How to use checksums
04b16adcd36aba77d33566030865e7383923b5d07958db7122855bac533916de
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.14

Release history Release notifications | RSS feed

This release

1.0.5 This release

2 release files

1.0.4

2 release files

1.0.3

2 release files

1.0.2

2 release files

1.0.1

2 release files

1.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page