The Fluency item type can be used to give your students pronunciation practice, with more useful feedback than simply that their speech was "right" or "wrong." You can try a sample activity here:

https://demo.poodll.com/mod/minilesson/view.php?id=1816


The following screenshot shows the key items on the page

  1. The progress dots - indicating success on 3 phrases/sentences so far
  2. The model audio AND student's audio for the current phrase/sentence.
  3. The current phrase / sentence the student is working on.
  4. The level of accuracy of the last attempt on the current phrase/sentence
  5. The last attempt marked up with green for "good", orange for "fair", and red for "not good".



The Fluency activity works in 20+ languages that are supported by Poodll. But it works best in English. 


Perfection is not required

It can be difficult for a student to get all "green" even when their pronunciation is good. If they have 5 yellow stars, or if they cannot get 5 yellow stars, they should continue. 


The threshold of pronunciation accuracy that is deemed "correct" can be configured in the item's "Correct Threshold" setting.

For non English languages, it is probably wise to lower it if the results are poor. This will make "green" words/letters easier to achieve and 5 yellow stars easier to get.


How is the star rating decided


Each word has an overall accuracy score (%). We use the average accuracy score of all the words in the result as the basis for the 5 star rating. Lets call that: scores_avg


star_bandwith: (100 - correct threshold) / 4
number. of stars = (scores_avg / star_bandwidth) + 1

That is effectively the calculation that is used, though the implementation might be slightly different. (e.g scores of 0 will actually get zero stars)


How is the color rating decided?


The accuracy score (acc_score)  for the entity we are considering (word / phoneme / syllable) is evaluated against the following scale to produce a color rating.

  • Green /  Correct: acc_score >= Correct threshold: 
  • Orange / Almost: acc_score >= (Correct threshold - 25)
  • Red / Incorrect:  acc_score < (Correct threshold - 25)
  • Grey: No recognition (not in results)


The "almost" threshold is arbitrarily set to: Correct threshold - 25. There is no option to configure that. 

But it is possible to not show "almost" at all (ie "almost" will be displayed as "incorrect" ). This is done in the Fluency item's instance setting: