add model card
Browse files
README.md
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language: aii
|
3 |
+
license: cc0-1.0
|
4 |
+
tags:
|
5 |
+
- audio
|
6 |
+
- automatic-speech-recognition
|
7 |
+
- speech
|
8 |
+
datasets:
|
9 |
+
- mnazari/urmi-assyrian-voice
|
10 |
+
metrics:
|
11 |
+
- cer
|
12 |
+
|
13 |
+
model-index:
|
14 |
+
- name: Wav2Vec2-Assyrian by Matthew Nazari
|
15 |
+
results:
|
16 |
+
- task:
|
17 |
+
type: automatic-speech-recognition
|
18 |
+
name: Speech Recognition
|
19 |
+
dataset:
|
20 |
+
type: mnazari/urmi-assyrian-voice
|
21 |
+
name: Urmi Assyrian Voice
|
22 |
+
metrics:
|
23 |
+
- type: cer
|
24 |
+
value: 7.5
|
25 |
+
name: Train CER
|
26 |
+
- type: cer
|
27 |
+
value: 11.7
|
28 |
+
name: Test CER
|
29 |
+
---
|
30 |
+
|
31 |
+
# Wav2Vec2-Assyrian
|
32 |
+
|
33 |
+
Fine-tuned [facebom3hrdadfiok/wav2vec2-large-xlsr-persian-v3](https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-persian-v3) in Assyrian (the Urmi dialect of North-Eastern Neo-Aramaic) using [Urmi Assyrian Voice](https://huggingface.co/datasets/mnazari/urmi-assyrian-voice). When using this model, note that certain characters need to be decoded:
|
34 |
+
|
35 |
+
| Symbol | Character |
|
36 |
+
| ------ | ---------- |
|
37 |
+
| 'α' | 'k̭' |
|
38 |
+
| 'β' | 'p̂' |
|
39 |
+
| 'γ' | 'č̭' |
|
40 |
+
| 'δ' | 'c̭' |
|