Few questions on the limitations of phrase list feature in Azure AI Speech

Question

Hi.

I was looking at the phrase-list feature in speech-text streaming flow and I have a few questions:

1. Is there a limit on the number of words we can provide as part of phraseListGrammar?
2. Is there a limit on the number of characters in each word?
3. Is there any feature for adding intensifiers? i.e., give more weight to the occurrence of a particular word.

https://learn.microsoft.com/en-gb/azure/ai-services/speech-service/improve-accuracy-phrase-list?tabs=terminal&pivots=programming-language-java

The above URL is what I'm referring to. There is no mention of limitations. Please point me to the right place if it's already documented.

Thanks,

Sai Vishnu Soudri

Answer 1

Hi @Sai Vishnu Soudri,

Thank you for reaching out to Microsoft Q&A forum!

Here’s the response to your queries:

Is there a limit on the number of words we can provide as part of phraseListGrammar?

There’s no explicit limit documented, but large phrase lists might cause performance issues. It’s best to test with your intended list size and consider breaking it into smaller chunks if needed.

Is there a limit on the number of characters in each word?

There's no documented limit on the number of characters per word in the phrase list grammar, but it’s good practice to keep words reasonably short to avoid unexpected behaviour or performance issues.

Is there any feature for adding intensifiers? i.e., give more weight to the occurrence of a particular word.

Azure Speech Service doesn’t directly support adding weight to specific words in the phrase list. However, you can improve the recognition accuracy for specific words by providing those words more frequently or in different contexts in your training data. Using a custom model or adjusting other aspects of the speech recognition pipeline might help you achieve the desired effect.

I hope you understand. Do let us know if you any further queries.

---

If this answers your query, do click Accept Answer and Yes for was this answer helpful.

Answer 2

The limit is 500. It is mentioned in the Introduction on the page.

You can use phrase lists with both standard and custom speech. There are some situations where training a custom model that includes phrases is likely the best option to improve accuracy. For example, in the following cases you would use custom speech:

• If you need to use a large list of phrases. A phrase list shouldn't have more than 500 phrases.
• If you need a phrase list for languages that aren't currently supported.