File size: 8,127 Bytes
507a0e8
b202bcd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
507a0e8
b202bcd
7bfc225
 
 
 
b202bcd
 
 
 
 
 
 
8d3063e
b202bcd
 
 
 
 
 
 
 
 
 
 
 
 
37664aa
b202bcd
 
 
 
 
 
 
 
37664aa
b202bcd
 
 
 
37664aa
b202bcd
 
 
 
 
 
 
 
 
 
 
 
 
37664aa
b202bcd
 
 
 
 
 
 
 
 
 
 
 
 
37664aa
b202bcd
 
 
 
 
 
 
 
 
507a0e8
b202bcd
 
 
 
 
 
 
 
 
 
 
 
 
 
3c99844
b202bcd
 
 
 
 
37664aa
b202bcd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5191a8f
 
 
5f243ee
3c99844
b202bcd
 
 
5191a8f
 
5f243ee
 
3c99844
b202bcd
 
 
 
 
 
 
 
 
 
 
3c99844
b202bcd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
---
language:
- en
library_name: nemo
datasets:
- VOXCELEB-1
- VOXCELEB-2
- FISHER
- switchboard
- librispeech_asr
- SRE (2004-2010) 
thumbnail: null
tags:
- speaker
- speech
- audio
- speaker-verification
- speaker-recognition
- speaker-diarization
- titanet
- NeMo
- pytorch
license: cc-by-4.0
widget:
- src: https://huggingface.co/nvidia/speakerverification_en_titanet_large/resolve/main/an255-fash-b.wav
  example_title: Speech sample 1
- src: https://huggingface.co/nvidia/speakerverification_en_titanet_large/resolve/main/cen7-fash-b.wav
  example_title: Speech sample 2
model-index:
- name: speakerverification_en_titanet_large
  results:
  - task:
      name: Speaker Verification
      type: speaker-verification
    dataset:
      name: voxceleb1
      type: voxceleb1-O
      config: clean
      split: test
      args:
        language: en
    metrics:
    - name: Test EER
      type: eer
      value: 0.66
  - task:
      type: Speaker Diarization
      name: speaker-diarization
    dataset:
      name: ami-mixheadset
      type: ami_diarization
      config: oracle-vad-known-number-of-speakers
      split: test
      args:
        language: en
    metrics:
    - name: Test DER
      type: der
      value: 1.73  
  - task:
      type: Speaker Diarization
      name: speaker-diarization
    dataset:
      name: ami-lapel
      type: ami_diarization
      config: oracle-vad-known-number-of-speakers
      split: test
      args:
        language: en
    metrics:
    - name: Test DER
      type: der
      value: 2.03
  - task:
      type: Speaker Diarization
      name: speaker-diarization
    dataset:
      name: ch109
      type: callhome_diarization
      config: oracle-vad-known-number-of-speakers
      split: test
      args:
        language: en
    metrics:
    - name: Test DER
      type: der
      value: 1.19
  - task:
      type: Speaker Diarization
      name: speaker-diarization
    dataset:
      name: nist-sre-2000
      type: nist-sre_diarization
      config: oracle-vad-known-number-of-speakers
      split: test
      args:
        language: en
    metrics:
    - name: Test DER
      type: der
      value: 6.73
---

# NVIDIA TitaNet-Large (en-US)

<style>
img {
 display: inline;
}
</style>

| [![Model architecture](https://img.shields.io/badge/Model_Arch-TitaNet--Large-lightgrey#model-badge)](#model-architecture)
| [![Model size](https://img.shields.io/badge/Params-23M-lightgrey#model-badge)](#model-architecture)
| [![Language](https://img.shields.io/badge/Language-en--US-lightgrey#model-badge)](#datasets)


This model extracts speaker embeddings from given speech, which is the backbone for speaker verification and diarization tasks.
It is a "large" version of TitaNet (around 23M parameters) models.  
See the [model architecture](#model-architecture) section and [NeMo documentation](https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/stable/asr/speaker_recognition/models.html#titanet) for complete architecture details.

## NVIDIA NeMo: Training

To train, fine-tune or play with the model you will need to install [NVIDIA NeMo](https://github.com/NVIDIA/NeMo). We recommend you install it after you've installed the latest Pytorch version.
```
pip install nemo_toolkit['all']
``` 

## How to Use this Model

The model is available for use in the NeMo toolkit [3] and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.

### Automatically instantiate the model

```python
import nemo.collections.asr as nemo_asr
speaker_model = nemo_asr.models.EncDecSpeakerLabelModel.from_pretrained("nvidia/speakerverification_en_titanet_large")
```

### Embedding Extraction

Using 

```python
emb = speaker_model.get_embedding("an255-fash-b.wav")
```

### Verifying two utterances (Speaker Verification)

Now to check if two audio files are from the same speaker or not, simply do:

```python
speaker_model.verify_speakers("an255-fash-b.wav","cen7-fash-b.wav")
```

### Extracting Embeddings for more audio files

To extract embeddings from a bunch of audio files:

Write audio files to a `manifest.json` file with lines as in format:

```json
{"audio_filepath": "<absolute path to dataset>/audio_file.wav", "duration": "duration of file in sec", "label": "speaker_id"}
```

Then running following script will extract embeddings and writes to current working directory:
```shell
python <NeMo_root>/examples/speaker_tasks/recognition/extract_speaker_embeddings.py --manifest=manifest.json
```

### Input

This model accepts 16000 KHz Mono-channel Audio (wav files) as input.

### Output

This model provides speaker embeddings for an audio file. 

## Model Architecture

TitaNet model is a depth-wise separable conv1D model [1] for Speaker Verification and diarization tasks. You may find more info on the detail of this model here: [TitaNet-Model](https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/main/asr/speaker_recognition/models.html). 

## Training

The NeMo toolkit [3] was used for training the models for over several hundred epochs. These model are trained with this [example script](https://github.com/NVIDIA/NeMo/blob/main/examples/speaker_tasks/recognition/speaker_reco.py) and this [base config](https://github.com/NVIDIA/NeMo/blob/main/examples/speaker_tasks/recognition/conf/titanet-large.yaml).

### Datasets

All the models in this collection are trained on a composite dataset comprising several thousand hours of English speech:

- Voxceleb-1
- Voxceleb-2
- Fisher
- Switchboard
- Librispeech
- SRE (2004-2010) 

## Performance

Performances of the these models are reported in terms of Equal Error Rate (EER%) on speaker verification evaluation trial files and as Diarization Error Rate (DER%) on diarization test sessions.

* Speaker Verification (EER%)
| Version | Model | Model Size | VoxCeleb1 (Cleaned trial file) |
|---------|--------------|-----|---------------|
| 1.10.0 | TitaNet-Large | 23M | 0.66   |

* Speaker Diarization (DER%)
| Version | Model | Model Size | Evaluation Condition | NIST SRE 2000 | AMI (Lapel) | AMI (MixHeadset) | CH109 |
|---------|--------------|-----|----------------------|---------------|-------------|------------------|-------|
| 1.10.0 | TitaNet-Large | 23M | Oracle VAD KNOWN # of Speakers  |      6.73     |      2.03      |         1.73        |  1.19 |
| 1.10.0 | TitaNet-Large | 23M | Oracle VAD UNKNOWN # of Speakers  |    5.38     |      2.03      |         1.89        |  1.63 |

## Limitations
This model is trained on both telephonic and non-telephonic speech from voxceleb datasets, Fisher and switch board. If your domain of data differs from trained data or doesnot show relatively good performance consider finetuning for that speech domain.

## NVIDIA Riva: Deployment

[NVIDIA Riva](https://developer.nvidia.com/riva), is an accelerated speech AI SDK deployable on-prem, in all clouds, multi-cloud, hybrid, on edge, and embedded. 
Additionally, Riva provides: 

* World-class out-of-the-box accuracy for the most common languages with model checkpoints trained on proprietary data with hundreds of thousands of GPU-compute hours 
* Best in class accuracy with run-time word boosting (e.g., brand and product names) and customization of acoustic model, language model, and inverse text normalization 
* Streaming speech recognition, Kubernetes compatible scaling, and enterprise-grade support 

Although this model isn’t supported yet by Riva, the [list of supported models is here](https://huggingface.co/models?other=Riva).  
Check out [Riva live demo](https://developer.nvidia.com/riva#demos). 

## References
[1] [TitaNet: Neural Model for Speaker Representation with 1D Depth-wise Separable convolutions and global context](https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9746806) 
[2] [NVIDIA NeMo Toolkit](https://github.com/NVIDIA/NeMo)

## Licence

License to use this model is covered by the [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/). By downloading the public and release version of the model, you accept the terms and conditions of the [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/) license.