A logistics company's voice ordering system misrecognizes spoken SKU codes and warehouse abbreviations. The team has recordings of staff saying these terms along with verified transcripts. What should they do to improve recognition of these terms?
Show answer and explanation
Answer: D
Custom Speech lets you train a speech-to-text model with your own audio paired with accurate transcripts, which teaches the recognizer the vocabulary and pronunciations specific to your business. Matched recordings and transcripts are the key ingredient. Glossaries apply to translation, semantic ranking improves search relevance, and opinion mining evaluates sentiment, so none of these improve how audio is recognized.
Why the other options are wrong
- A. A translation glossary controls how specific words are rendered in translated output. It does not change how spoken SKU codes are recognized from audio.
- B. Semantic ranking reorders search results using language models. It operates after text has been produced and does not help the recognizer hear the codes correctly.
- C. Opinion mining identifies sentiment toward aspects of text. It does not improve transcription of the spoken terms in the first place.