Downloads · 30 days
0
aisyahsyihab/voxcpm-lora-indonesian-female
voxcpm-lora-indonesian-female is a text-to-speech model from aisyahsyihab. Use it when you need text read aloud. It is set up for voxcpm. The card lists the license as apache-2.0.
LoRA adapter for VoxCPM 1.5 fine-tuned for natural Indonesian female speech.
Downloads · 30 days
0
Access
Public
Updated Mar 19, 2026
Repo size
32.8 MB
Likes
0
Public
Click a slice to open those files.
.safetensors32.8 MB · 100%
From the Hugging Face model README
LoRA adapter for VoxCPM 1.5 fine-tuned for natural Indonesian female speech.
| Property | Value |
|---|---|
| Base model | openbmb/VoxCPM1.5 (800M params) |
| LoRA rank (r) | 32 |
| LoRA alpha | 16 |
| Training steps | 3500 |
| Training samples | 112 clips |
| Final loss | 0.790 |
| Sample rate | 44.1 kHz |
Safetensors support requires the latest VoxCPM from GitHub:
pip install git+https://github.com/openbmb/VoxCPM.git
from voxcpm import VoxCPM
from huggingface_hub import snapshot_download
import soundfile as sf
# Download LoRA
lora_path = snapshot_download("aisyahsyihab/voxcpm-lora-indonesian-female")
# Load model with LoRA
model = VoxCPM.from_pretrained(
"openbmb/VoxCPM1.5",
lora_weights_path=lora_path,
load_denoiser=False
)
# Generate speech
audio = model.generate(
text="Halo, apa kabar hari ini?",
cfg_value=3.0,
normalize=True
)
# Save output
sf.write("output.wav", audio, model.tts_model.sample_rate)
lora:
enable_lm: true
enable_dit: true
enable_proj: false
r: 32
alpha: 16
dropout: 0.0
target_modules_lm: ["q_proj", "v_proj", "k_proj", "o_proj"]
target_modules_dit: ["q_proj", "v_proj", "k_proj", "o_proj"]
training:
batch_size: 2
grad_accum_steps: 4
learning_rate: 0.0001
warmup_steps: 100
max_steps: 4000
Apache 2.0 (following VoxCPM base model license)