JavaScript required
We’re sorry, but Coda doesn’t work properly without JavaScript enabled.
Gallery
AI4Bharat
Share
Explore
Gallery
AI4Bharat
AI4Bharat Public
Seminars
Publications
People
Models
AI4Bharat Admin
Members
Planning
Licensing
Meity Timelines
Hiring
AI4Bharat Summer of Code
IndicMining
Meeting Minutes
NeurIPS dataset paper plan
IndicASR
RNN-T
Multilingual ASR
Analysis
Adaptation in End-to-End Speech Recognition
Data Augmentation
Text Normalization for speech
Shoonya
Documentation - User Manual
Welcome Page
User-Roles on Shoonya
Getting Started with Workflow
Manager Workflow
Language-Experts Workflow
Annotation Workflow
Collection Workflow
Terminology
FAQs and Feedback
Management Dashboard
Language Experts
Annotation Tasks
Reporting and Analytics
Projects DataExports
Task Details
Shoonya Development Document
Shoonya Workflow
Software Architecture Diagrams
Technology Used
Shoonya Code Structure
Shoonya Deployment
Shoonya Forms
Feature Suggestions
Report Bugs for Shoonya
User Feedbacks
Stats-collection Forms
Multilingual ASR
Analysis
Model Name:
Multisoftmax Variants
Multisoftmax Variants
3
Model Name
Language
odia
9
bengali
9
telugu
12
gujarati
12
hindi
9
marathi
9
tamil
12
Test set
dcunk_new
3
dckn_new
3
mucs
3
dcunk_new
3
dckn_new
3
openslr
3
dcunk_new
3
dckn_new
3
mucs
3
msr
3
dcunk_new
3
dckn_new
3
mucs
3
msr
3
dcunk_new
3
dckn_new
3
mucs
3
dcunk_new
3
dckn_new
3
mucs
3
dcunk_new
3
dckn_new
3
mucs
3
msr
3
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
Model Name
Language
odia
9
bengali
9
telugu
12
gujarati
12
hindi
9
marathi
9
tamil
12
Test set
dcunk_new
3
dckn_new
3
mucs
3
dcunk_new
3
dckn_new
3
openslr
3
dcunk_new
3
dckn_new
3
mucs
3
msr
3
dcunk_new
3
dckn_new
3
mucs
3
msr
3
dcunk_new
3
dckn_new
3
mucs
3
dcunk_new
3
dckn_new
3
mucs
3
dcunk_new
3
dckn_new
3
mucs
3
msr
3
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
base_dc_unnormalized
24
26.83
6.98
22.88
6.1
37.65
7.9
20.1
5.72
19.19
5.54
74.63
18.21
29.18
5.32
25.4
4.09
41.55
10.11
33.92
7.43
18.01
4.8
18.32
4.93
42.25
15.05
35.18
11.44
12.64
3.93
10.57
3.29
30.67
11.46
22.88
6.06
20.87
5.16
35.41
6.65
27.57
4.29
26.35
4.13
35.34
7.9
29.89
5.98
base_dc_multisoftmax_with_lid
24
26.76
6.98
23.07
6.16
37.61
7.84
20.39
5.84
19.68
5.66
88.01
16.93
29.47
5.47
25.97
4.28
42.1
10.21
34.45
7.47
18.01
4.85
18.6
5.01
42.52
15.12
35.53
11.58
12.69
3.97
10.92
3.4
30.22
11.29
23.12
6.16
21.34
5.34
37.58
6.2
27.48
4.32
26.65
4.22
35.74
8.05
30.17
6.11
base_dc_embedding_768_3layer
24
26.48
6.91
22.94
6.08
35.89
7.21
19.96
5.72
19.18
5.52
85.28
17.29
29.01
5.37
25.3
4.12
41.25
9.86
33.89
7.31
17.77
4.78
18.25
4.95
42.29
15.09
35.15
11.51
12.29
3.84
10.62
3.29
30.06
11.24
22.65
6.01
20.93
5.2
35.28
6.2
27.12
4.31
26.29
4.12
35.6
7.95
29.89
6.01
Effect of batch size
Effect of batch size
3
Model Name
Language
odia
6
bengali
6
telugu
8
gujarati
8
hindi
6
marathi
6
tamil
8
Test set
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
openslr
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
Model Name
Language
odia
6
bengali
6
telugu
8
gujarati
8
hindi
6
marathi
6
tamil
8
Test set
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
openslr
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
base_dc_embedding_768_3layer
24
26.48
6.91
22.94
6.08
35.89
7.21
19.96
5.72
19.18
5.52
85.28
17.29
29.01
5.37
25.3
4.12
41.25
9.86
33.89
7.31
17.77
4.78
18.25
4.95
42.29
15.09
35.15
11.51
12.29
3.84
10.62
3.29
30.06
11.24
22.65
6.01
20.93
5.2
35.28
6.2
27.12
4.31
26.29
4.12
35.6
7.95
29.89
6.01
base_dc_embedding_768_3layer_bs2x
24
25.93
6.7
21.93
5.85
36.05
7.1
18.75
5.45
17.79
5.24
88.93
19.18
26.87
4.83
23.65
3.73
39.95
9.45
32.43
6.96
16.63
4.45
16.9
4.55
40.61
14.48
33.71
10.92
11.32
3.57
9.7
2.99
28.92
10.85
21.11
5.68
19.37
4.82
34.89
6.29
25.84
3.95
24.35
3.7
34.12
7.57
28.63
5.72
Effect of normalization
Effect of normalization
3
Model Name
Language
odia
6
bengali
6
telugu
8
gujarati
8
hindi
6
marathi
6
tamil
8
Test set
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
openslr
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
mucs
2
dcunk_new
2
dckn_new
2
mucs
2
msr
2
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER
CER
WER