Update recipe for linguistic model

Hi,

Please update the readme and sourcecode for training the linguistic model.

I assumed I had to go inside the data_loader and execute generate_transcriptions . Also, I had to download the deepspeech models after I figure out I need them, ~~also I think the deepspeech_generator is deprecated, so I had to make a sys call~~

Thanks again!

I found out that the "deep_generator" was a script on the speech_emotion_recognition folder. My bad. Anyway the steps for generating the linguistic models will be appreciated :). Thanks again.

Managed generate ASR transcriptions run it by running:

$ python
import speech_emotion_recognition.data_loader as dl
from speech_emotion_recognition import *
dl.generate_transcriptions()

I hava download deepspeech model.

Deepspeech 0.5.1 must use ubuntu. Then the generate_transcriptions() will be worked .

But a new problem is here:
How to get the file of embeddings_array.numpy and word_to_index.pickle?

speech-emotion-recognition/speech_emotion_recognition/word2vec_wrapper.py

Lines 14 to 15 in 8ecbcf1

    
           cls.embedding_array = np.load("data/embeddings_array.numpy") 
        
           cls.word_to_index = pickle.load(open("data/word_to_index.pickle", "rb"))

@PiotrSobczak

Hello, I would like to ask where to generate the NPY file

the same question about how to generate embeddings_array.numpy and word_to_index.pickle。。。Thanks

	cls.embedding_array = np.load("data/embeddings_array.numpy")
	cls.word_to_index = pickle.load(open("data/word_to_index.pickle", "rb"))