Downloads · 30 days
4
15% of all-time downloads
fast-conformer-id-jv
Isaacyn/fast-conformer-id-jv
fast-conformer-id-jv is a automatic speech recognition model from Isaacyn. Use it when you need speech turned into text. It is set up for nemo. The card lists the license as cc-by-4.0.
Fine-tuned Fast Conformer dengan LoRA untuk ASR tiga bahasa: Bahasa Indonesia, Bahasa Jawa, dan Bahasa Inggris. Output selalu Bahasa Indonesia.
All-time downloads
26
Public
Repo size
947 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.onnx483 MB · 51%
At a glance
- Task
- Automatic Speech Recognition
- Library
- nemo
- License
- cc-by-4.0
- Access
- Public
- Created
- May 17, 2026
- Updated
- May 17, 2026
- SHA
- 16df54ba
Model card
From the Hugging Face model README
Fast Conformer ASR — Bahasa Indonesia, Bahasa Jawa & Bahasa Inggris
Fine-tuned Fast Conformer dengan LoRA untuk ASR tiga bahasa: Bahasa Indonesia, Bahasa Jawa, dan Bahasa Inggris. Output selalu Bahasa Indonesia.
Model Details
- Base model: Fast Conformer (Rekesh et al., 2023)
- Fine-tuning: LoRA/PEFT via HuggingFace PEFT
- Languages input: Bahasa Indonesia (
id), Bahasa Jawa (jv), Bahasa Inggris (en) - Output: Bahasa Indonesia
- Framework: NVIDIA NeMo
- Export: ONNX Runtime compatible
Files
fast_conformer_multilingual.nemo— NeMo checkpointfast_conformer_multilingual.onnx— ONNX export untuk inferensi/browser extension
Datasets
- Mozilla Common Voice 17.0 (id, CC0-1.0)
- Google FLEURS (id_id, jv_jv, CC-BY 4.0)
- OpenSLR SLR35 — Large Javanese ASR Training Dataset
- LibriSpeech (en)
Citation
Rekesh et al. (2023) — Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition.
- Task
- Automatic Speech Recognition
- Library
- nemo
- License
- cc-by-4.0
- Languages
- id, jv, en
- Created
- May 17, 2026
- Updated
- May 17, 2026