Natural Language Processing for Speech and Text: From Beginner to Advanced

Course details

With the recent surge in large language models, it's particularly relevant to explore the evolution of NLP techniques, from traditional methods to current industry standards. In this course, Wuraola Oyewusi—an experienced data scientist and machine learning and artificial intelligence professional—helps you build a strong foundation in natural language processing (NLP) concepts and addresses the end-to-end application of NLP. Learn about both text and speech data while you explore the theoretical background of NLP concepts, the historical evolution of NLP techniques, and current applications of NLP representation techniques for both text and speech data. Dive into code-based practice exercises for preprocessing techniques and tasks for both text and speech data. Plus, check out a wide range of Python libraries, including NLTK, spaCy, Hugging Face, Transformers, librosa, scikit-learn, gensim, and torchaudio.

Learning objectives
Understand the functions and evolution of natural language processing for speech and text data (machine learning-based and other techniques).
Implement in Python the techniques that are taught.
Get started with NLP and fill in important knowledge gaps.

Concepts

0. Introduction

01 - Fundamentals of natural language processing
02 - NLP course strategy

1. Introduction to Natural Language Processing (NLP)

03 - What is natural language processing (NLP)
04 - What are sequences
05 - Applications of natural language processing in text data
06 - Applications of natural language processing in speech data
07 - Historical evolution of NLP tasks and techniques
08 - How computers understand sequences in NLP

2. Natural Language Processing for Text Techniques

09 - Text preprocessing
10 - Text preprocessing using NLTK
11 - Text representation
12 - Text representation - One-hot encoding
13 - One-hot encoding using scikit-learn
14 - Text representation - N-grams
15 - N-grams representation using NLTK
16 - Text representation - Bag-of-words (BoW)
17 - Bag-of-words representation using scikit-learn
18 - Text representation - Term frequency-inverse document frequency (TF-IDF)
19 - TF-IDF representation using scikit-learn
20 - Text representation - Word embeddings
21 - Word2vec embedding using Gensim
22 - Embedding with pretrained spaCy model
23 - Sentence embedding using the Sentence Transformers library
24 - Text representation - Pre-trained language models (PLMs)
25 - Pre-trained language models using Transformers

3. Natural Language Processing for Speech Techniques

26 - Speech representation - Mel-frequency cepstral coefficients
27 - Mel-frequency cepstral coefficients (MFCCs) using librosa
28 - Speech representation - Linear predictive cepstral coefficients (LPCCs)
29 - Linear predictive coding (LPC) using librosa
30 - Speech representation - Gammatone filterbank features
31 - Gammatone filterbank features using librosa
32 - Speech representation - Spectrograms
33 - Spectrograms using fast Fourier transform (FFT) in librosa
34 - Speech representation - Speech embeddings
35 - Speech embeddings using wav2vec in Transformers

4. Applied Natural Language Processing - Algorithms and Tasks

36 - Algorithms for natural language processing tasks
37 - Types of algorithms in natural language processing
38 - Rule-based - Regular expressions
39 - Regular expression tasks using the re library
40 - Rule-based - Rule-based parsing
41 - Parsing sentences into syntactic structures using context-free grammars (CFG) in NLTK
42 - Part-of-speech (POS) tagging using spaCy
43 - Statistical - Hidden Markov models (HMMs)
44 - Hidden Markov models (HMMs) for POS tagging in NLTK
45 - Statistical - Conditional random fields (CRFs)
46 - Statistical - Naive Bayes classifiers
47 - Machine learning - Support vector machines (SVMs)
48 - Classify text data using SVM
49 - Machine learning - Decision trees
50 - Classify the speech commands dataset using decision trees
51 - Machine learning - K-means clustering
52 - K-means clustering for the movie reviews dataset
53 - Deep learning - Recurrent neural networks (RNNs)
54 - Text generation using recurrent neural networks (RNNs)
55 - Deep learning - Transformers
56 - Transfer learning in natural language processing (NLP)
57 - Speech-to-text (STT) using wav2vec in the Transformers library
58 - Text-to-speech (TTS) using Tacotron and WaveGlow

Conclusion

59 - What's next - NLP in practice

Natural Language Processing for Speech and Text: From Beginner to Advanced

Authors

Wuraola Oyewusi