File size: 1,217 Bytes
0cf5f14
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
74
75
76
77
78
---
language:
- akp
- ada
- acd
- any
- avn
- bib
- bim
- biv
- bov
- bud
- bwu
- dag
- dga
- en
- ewe
- fat
- ffm
- gjn
- gur
- hau
- kbp
- kdh
- kma
- kus
- lef
- lip
- maw
- mzw
- naw
- ncu
- nko
- ntr
- nzi
- sfw
- sig
- sil
- snw
- tpm
- vag
- xon
- xsm
language_bcp47:
- twi-akuapem
- twi-asante
license: cc-by-nc-4.0
library_name: voxcpm
tags:
- voxcpm
- ghana-tts
- text-to-speech
- ghanaian-languages
pipeline_tag: text-to-speech
---

# Ghana TTS – 36k

Checkpoint at **36,000 steps** for the Ghana TTS model, fine-tuned from [VoxCPM](https://huggingface.co/VoxCrimson/voxcpm-0.5B) on 1,400+ hours of Ghanaian language speech data across 44 languages.

Trained on a single NVIDIA H200 (140 GB VRAM) using bf16 + TF32.

## Contents

- `model.safetensors` – model weights
- `optimizer.pth` – optimizer state
- `audiovae/` – audio VAE components
- `scheduler/` – noise scheduler config
- `tokenizer/` – tokenizer files
- `prompt_audio/` – reference audio samples for voice prompting

## Usage

See the [Ghana TTS space](https://huggingface.co/spaces/ghananlpcommunity/ghana-tts) for inference.

## Training

Training code: [github.com/michsethowusu/VoxCPM](https://github.com/michsethowusu/VoxCPM)