docs: add STT model card and direct endpoints documentation
Browse files
README.md
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
language:
|
| 4 |
+
- hi
|
| 5 |
+
- bn
|
| 6 |
+
- te
|
| 7 |
+
- ta
|
| 8 |
+
- mr
|
| 9 |
+
- gu
|
| 10 |
+
- kn
|
| 11 |
+
- ml
|
| 12 |
+
- pa
|
| 13 |
+
- or
|
| 14 |
+
- as
|
| 15 |
+
- ur
|
| 16 |
+
- sa
|
| 17 |
+
- ne
|
| 18 |
+
- kok
|
| 19 |
+
- mai
|
| 20 |
+
- brx
|
| 21 |
+
- doi
|
| 22 |
+
- ks
|
| 23 |
+
- mni
|
| 24 |
+
- sat
|
| 25 |
+
- sd
|
| 26 |
+
- en
|
| 27 |
+
tags:
|
| 28 |
+
- audio
|
| 29 |
+
- automatic-speech-recognition
|
| 30 |
+
- speech-to-text
|
| 31 |
+
- onnx
|
| 32 |
+
- sherpa-onnx
|
| 33 |
+
- indic
|
| 34 |
+
- indicconformer
|
| 35 |
+
pipeline_tag: automatic-speech-recognition
|
| 36 |
+
---
|
| 37 |
+
|
| 38 |
+
# BharatBhasha AI - Speech-to-Text (STT) Models
|
| 39 |
+
|
| 40 |
+
This repository hosts quantized ONNX INT8 Conformer-CTC Automatic Speech Recognition (ASR) models for **all 22 Official Scheduled Indian Languages + English**, powered by AI4Bharat and Sherpa-ONNX.
|
| 41 |
+
|
| 42 |
+
> [!NOTE]
|
| 43 |
+
> **Application Distribution Mirror**:
|
| 44 |
+
> Maintained specifically as an un-gated distribution endpoint for client applications (Android, Desktop) to ensure continuous availability and resilience against upstream network/hosting failures.
|
| 45 |
+
> All original model weights, research, and architectures are credited to **AI4Bharat (IIT Madras)** and the **Sherpa-ONNX (Next-gen Kaldi)** community under the **MIT / Apache 2.0 Licenses**.
|
| 46 |
+
|
| 47 |
+
## Direct Download Endpoint
|
| 48 |
+
```text
|
| 49 |
+
# Acoustic Model
|
| 50 |
+
https://huggingface.co/remiai3/STT_MODELS_FOR_APP/resolve/main/indic_conformer_{LANG_CODE}.onnx
|
| 51 |
+
|
| 52 |
+
# Vocabulary Tokens
|
| 53 |
+
https://huggingface.co/remiai3/STT_MODELS_FOR_APP/resolve/main/tokens.txt
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
## Supported Languages (23 Total)
|
| 57 |
+
- Hindi (`hi`), Bengali (`bn`), Telugu (`te`), Tamil (`ta`), Marathi (`mr`), Gujarati (`gu`)
|
| 58 |
+
- Kannada (`kn`), Malayalam (`ml`), Punjabi (`pa`), Odia (`or`), Assamese (`as`), Urdu (`ur`)
|
| 59 |
+
- Sanskrit (`sa`), Nepali (`ne`), Konkani (`kok`), Maithili (`mai`), Bodo (`brx`), Dogri (`doi`)
|
| 60 |
+
- Kashmiri (`ks`), Manipuri (`mni`), Santali (`sat`), Sindhi (`sd`), English (`en`)
|