@@ -8,7 +8,7 @@ To illustrate the model implementation clearly, 3 parts are described in detail.
- Data Preparation
- Encoder
- Decoder
In addition, the training process and the testing process are also introduced.
The arcitecture of the model is shown in Fig.1.
...
...
@@ -145,4 +145,4 @@ After the training process, we use stage 3,4,5 for testing process. The stage 3
## No Streaming
The deepspeech2 offline model is similarity to the deepspeech2 online model. The main difference between them is the offline model use the bi-directional rnn layers while the online model use the single direction rnn layers.
The deepspeech2 offline model is similarity to the deepspeech2 online model. The main difference between them is the offline model use the bi-directional rnn layers while the online model use the single direction rnn layers. The arcitecture of the model is shown in Fig.2.
