Common Error Labels for Volunteer Audio Clips
In a speech-to-text training set built from volunteer recordings, you can count recurring problems such as loud background noise, speaking too quickly, and recording from too far away from the microphone.
0
1
Tags
Machine Learning
Deep Learning
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Strategy
Machine Learning Yearning @ DeepLearning.AI
Related
Common Error Labels for Volunteer Audio Clips
What should you do first when a model is making many mistakes on the training set?
True or False: A useful way to inspect training mistakes is to group them into categories and count how often each type appears, similar to dev-set error analysis.
How many bad training examples should you inspect first?
Match Each Training-Review Idea to Its Description
Put the training-set review process in the correct order.
Why inspecting a small set of bad training cases can reveal training-set problems
Diagnosing Poor Training Performance in a Speech Model
What should you do when training performance is poor?
What is the main purpose of reviewing training examples that the model handles poorly?
True or False: Diagnosing model failures requires checking every training example in the dataset.
Learn After
Which labeled error category fits a recording with another voice talking over the main speaker?
Speech Review Error Categories
Complete the category: Too far from the _____.
Match each review label to the condition it records in a voicemail sample.
Order the steps for reviewing a low-quality field recording.
How Label Categories Organize a Review of Weak Call Recordings
Decide which recording issues apply to a distant, noisy audio sample.
What three conditions can be counted when reviewing these call recordings?
Which label fits a recording where the speaker talks too fast, with no sign of noise or microphone-distance problems?
A single speech sample can be labeled for both reverberation and clipping if both defects are present.