Read Aloud makes life easy for teachers by transcribing the student audio and matching it to the passage for you. But it still makes mistakes. So you will want to "tune" your passage to minimize mis-transcriptions. In most cases you will do this by specifying "alternatives" which acceptable mistranscriptions. In some cases you might choose to edit your passage to be easier to transcribe. In that case, after editing you should delete all attempts and start from scratch. Here we will be dealing with tuning your activity by creating a list of "alternatives."
There are two ways to see how AI is transcribing your readings, via an attempt's grading page and via the Alternatives Admin tab. In both cases you will want to have a few readings of the passage completed, so that you have transcriptions to work with.
Using the Attempt Grading Page
- Go into "grading" tab after ReadAloud has processed an attempt.
- Tap the attempt's "Grade Now" button
- Tap the "Transcript" button to set transcript mode
- Tap on the "error" words to see what ReadAloud thought you said
- Make a list of the mis-transcriptions
Using the Alternatives Admin tab
- Go into "Alternatives Admin" tab after some readings have been processed
- Review the Summary of Mistranscriptions table for acceptable mistranscriptions
- Enter acceptable mistranscriptions in the "Alternatives" field
- Tap "Save and Re-evaluate"
Each ReadAloud activity has an "Alternatives" area. This is where you can specify which mis-transcriptions you will accept for the specified passage word.
|Passage Word||Acceptable transcriptions|
In the alternatives text area the word sets above would be listed each in their own separated by the | character. The first word is the "key", ie the passage word. The subsequent ones are transcriptions.
NB You do not need to worry about differences between word case or punctuation. In fact is best to skip punctuation when writing alternatives.
For each common or likely mis-transcription make an alternatives entry.
You can not use multi word phrases as passage words or as acceptable transcriptions. Just individual words.
You will need to specify forward matches in a special but common case. That is when a single transcript word needs to match a two word phrase in the passage.
e.g AI will often mis-transcribe "you are" as "you're" or "your"
But we can not specify multi word phrases(eg "you are") as keys in an alternatives set. In the example case we could set "you" to "your" but then the second word "are" will not be matched by anything. To deal with this we can add a forward match before the closing pipe of the acceptable transcription.
So the solution to the above examples, is below. This will mark both "you" and "are" as correct if "you're" or "your" is transcribed
you|youre are|your are
You can use a wildcard as an acceptable transcription. Though you can not use forward matches with wildcards.
Wildcards are applied after all the other matching is done. Any unmatched wildcard words will then be flagged as matched. So this would work: a|along long|*
Because the wildcard would not be matched until any other matching directives were applied.
NB You can not use forward matches with wildcards.
The Final Syntax
So the final syntax looks like this:
passage-word | OK-transcription forward-match(optional) | OK-transcription forward-match(optional) | ...
The following list might be useful as a start place base for English Read Aloud passages.
|Passage word||Alternative words||Full string|
|I||im am|ive have||i|im am|ive have|
|you||your|youre are||you|your|youre are|
|he||hes is||he|hes is|
|she||shes is||she|shes is|
|they||theyre are||they|theyre are|
|there||theres is||there|theres is|
|we||were are|well will||we|were are|well will|
For all you copy and paste'rs you can copy this:
i|im am|ive have you|your|youre are he|hes is she|shes is they|theyre are there|theres is theyre|there|their their|their|theyre then|than of|off we|were are|well will