Published Mar 8, 2019

Human vs Machine Transcription

Explore the fascinating challenges and advancements in speech recognition technology as hosts dive into the battle of human vs. machine transcription, examining key benchmarks and uncovering the shared obstacles of transcription errors influenced by accents and audio quality.

Episode Highlights

Topics covered

Episode Highlights

Related Episodes

Chatbots
Answers 383 questions
The Limits of NLP
Answers 383 questions
NLP for Developers
Answers 383 questions
Interpretability Practitioners
Answers 383 questions
AI Roundtable
Answers 383 questions
Annotator Bias
Answers 383 questions
Computer-based Personality Judgments
Answers 383 questions
Machine Learning on Images with Noisy Human-centric Labels
Answers 383 questions
Let's Talk About Natural Language Processing
Answers 383 questions
Machine Learning Done Wrong
Answers 383 questions
The Transformer
Answers 383 questions
Interpretability Tooling
Answers 383 questions
Evolutionary Computation
Answers 383 questions
Brain Inspired AI
Answers 383 questions
Transfer Learning
Answers 383 questions

Dexa/Kyle Polich

Human vs Machine Transcription

Topics covered

Popular Clips

Misrecognized Words

Surprising Machine Errors

Switchboard Corpus Analysis

Machine vs Human Transcription

Human Level Transcription

Advancements in Speech Recognition

Human vs Machine Accuracy

Advancements in Speech Recognition

Speech Recognition Insights

Speech Recognition Insights

Episode Highlights

Speech Recognition Advances

Benchmark Speech Tasks

Human vs Machine

Related Episodes

Chatbots

The Limits of NLP

NLP for Developers

Interpretability Practitioners

AI Roundtable

Annotator Bias

Computer-based Personality Judgments

Machine Learning on Images with Noisy Human-centric Labels

Let's Talk About Natural Language Processing

Machine Learning Done Wrong

The Transformer

Interpretability Tooling

Evolutionary Computation

Brain Inspired AI

Transfer Learning

Human vs Machine Transcription

Topics covered

Popular Clips

Episode Highlights

Speech Recognition AdvancesThe episode explores the current state and future prospects of speech recognition technology, highlighting its reliance on neural networks and the challenges in surpassing human transcription capabilities.

Speech Recognition Advances

Benchmark Speech TasksThe Switchboard corpus is a crucial tool for testing speech recognition systems, providing a benchmark for comparing human and machine transcription accuracy. Insights reveal that while both face similar challenges, machines have the advantage of scalability without fatigue.

Benchmark Speech Tasks

Human vs Machine

Related Episodes