With existing free sites like Speakable and ReadM, there’s not a great need to use tools like ChatGPT, Bard, or Poe on their own to help ELLs with pronunciation.

Nevertheless, I figured it couldn’t hurt to experiment with Poe, which has a microphone option on its web version (ChatGPT just has a mike on its mobile app), to see if it might offer some unique advantages.

As you can see from the screenshot above, this was the prompt I gave it:

I am going to record the following paragraph. I am learning English. Please note where I do well and where I mispronounce a word. Reproduce the corrected paragraph and put those mispronounced words in bold:

After that, I pasted a few sentences, and then began saying them.    Poe caught most, though not all, of my deliberate mispronunciations.

I could see this kind of practice could be helpful to ELLs when they were preparing for a presentation, or just when they wanted to have a little more agency over the process and choose their own passages to practice (In Speakable and ReadM, users need to practice with predetermined exercises).

I’m adding this info to My Best Posts On Artificial Intelligence In The Classroom.