Skip to content

Incorrect pronunciation of "convert" using Polly's generative engine

0

Input Text

Convert 341 L 756 mL to milliliters.

Region

us-east-1

Issue

"Convert" is pronounced like the noun (/ˈkänˌvərt/) instead of the verb (/kənˈvərt/).

Steps to Reproduce

  1. Go to Polly text-to-speech console on the web: https://console.aws.amazon.com/polly/home/SynthesizeSpeech.
  2. Paste in "Convert 341 L 756 mL to milliliters."
  3. Choose Generative engine, US, English language, and Ruth, Female voice.
  4. Hear "convert" pronounced like the noun.

If you try it in the neural engine, it is pronounced correctly. I did not test all voices, but upon spot-check, it appears to be an issue across voices.

Ideal Solution

Using lexicons, SSML, or switching engines just for this sentence is possible but feels excessive for our workflow. Can this be corrected in the generative engine's pronunciation model?

asked a year ago133 views

1 Answer
0

Thank you for reporting this!

We are currently working on improving the pronunciation of that use case. In the meantime, you can change the pronunciation in your input text to the desired one by using the SSML tag "phoneme". See the example below:

<speak><phoneme alphabet='x-sampa' ph='k@n.\"v3`t'>Convert</phoneme> 341 L 756 mL to milliliters.</speak>

Sources
Improving pronunciation by specifying parts of speech - Amazon Polly
Managing lexicons - Amazon Polly
Neural voices - Amazon Polly

answered a year ago

AWS
SUPPORT ENGINEER

revised a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.