Essay

Explain the cause of poor performance in the in-car speech recognition scenario.

Question: In the context of the in-car audio data mismatch example, analyze why a speech recognition system might exhibit significantly degraded performance when evaluated on the dev set compared to the training set. Discuss the specific environmental factors involved.

Sample answer: The speech recognition system's performance degrades because the dev set contains audio clips recorded within a car, which introduces engine and road noise. In contrast, the training set consisted mostly of examples recorded against a quiet background. This stark difference in the acoustic environment creates a data mismatch, meaning the model is not prepared for the noisy conditions found in the dev set.

Key points:

  • Training set recorded against a quiet background
  • Dev set recorded inside a car
  • Engine and road noise worsen performance

Rubric: A full-credit answer must identify the difference in recording environments (quiet vs. inside a car) and explicitly mention the presence of engine and road noise as the cause for the performance drop.

0

1

Updated 2026-06-13

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Machine Learning Yearning @ DeepLearning.AI