🗝 KeyzHub
19Keys · community archive
2403 bytes raw 
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
# Windows and Linux appendix

The core workflow in the main README (Ollama chat, the local API, model sizing by RAM,
and wiring into agent tools via the OpenAI-compatible endpoint) is identical across
platforms once Ollama is installed. This page covers the platform-specific install
and voice-stack steps.

## Install Ollama

### Windows

Download the installer from [ollama.com/download](https://ollama.com/download) and run it.
Ollama installs as a background service and starts automatically. Verify it's running:

```powershell
ollama run llama3.2
```

### Linux

```bash
curl -fsSL https://ollama.com/install.sh | sh
```

This installs Ollama and sets it up as a systemd service. Check status with:

```bash
systemctl status ollama
```

Once installed, everything in README.md sections 1, 2, and 5 (pulling models, chatting,
choosing sizes by RAM, and the OpenAI-compatible API wiring) works exactly the same as
on macOS — Ollama's API is identical across platforms.

## LM Studio

LM Studio ships native installers for Windows and Linux at [lmstudio.ai](https://lmstudio.ai).
The app and local server work the same way described in README.md section 3.

## Local voice on Windows/Linux

The Mac guide's voice section (README.md section 4) uses Metal acceleration, which is
Apple-only. On Windows and Linux:

### Speech-to-text (whisper.cpp)

Build from source or use prebuilt binaries from the
[whisper.cpp releases page](https://github.com/ggml-org/whisper.cpp/releases). On Linux
you can also build with CUDA support if you have an NVIDIA GPU, for much faster
transcription:

```bash
git clone https://github.com/ggml-org/whisper.cpp
cd whisper.cpp
cmake -B build -DGGML_CUDA=1   # omit -DGGML_CUDA=1 if you have no NVIDIA GPU
cmake --build build --config Release
```

### Text-to-speech (Chatterbox)

mlx-audio is Apple MLX-specific and will not run on Windows/Linux. Use the upstream
[Chatterbox](https://github.com/resemble-ai/chatterbox) repository directly instead,
which runs on PyTorch and supports NVIDIA GPUs via CUDA:

```bash
python3 -m venv chatterbox-env
source chatterbox-env/bin/activate    # On Windows: chatterbox-env\Scripts\activate
pip install chatterbox-tts
```

Follow the usage examples in the Chatterbox repository README for generation and
voice-cloning commands — the CLI shape is similar to the mlx-audio commands in the
main README, but flags differ by version.