Text-to-speech mispronounced words

Hi Will,

Do it phonetically in the notes. In this case, "prezents" ends up sounding pretty good.

Thanks David. I've tried regenerating the audio using "prezents" instead of "presents" in that sentence (using the voice named Joanna or Matthew), but I'm not hearing much of a difference, unfortunately.

When looking up a possible solution, I noticed that Captivate has this option for adding words to a dictionary... https://elearning.adobe.com/2018/08/words-pronounced-incorrectly-text-speech/

So I was wondering if there is something similar. All-in-all, I do think the text-to-speech voices in Storyline are excellent though!

I hear a difference, but maybe it has to do with the specific voice chosen (I used Joanna).

I tried it a different way, too:

I received many great birthday presents.

I like how this speaker prezents her material.

I like how this speaker pre zents her material.

Personally, I think "prezents" sounds like I would expect.

Hi David,

I think the problem is that you are not using the exact sentence I mentioned. Try the phrase "The patient presents to a hospital."

Joining the discussion. I also have issues with how certain words using text-to-speech sound. I'm hoping someone came up with a cool trick. 

Type the text like this:

"The patient preZents to a hospital’s dedicated emergency department and requests care for a medical condition."

The capital "Z" seems to make the difference.

Thanks Ben! Capitalizing the "Z" worked! So between David's suggest of 'z' and your suggestion of capitalizing it, this is now passable! I wonder if capitalizing a letter in the middle of a word always shifts the accent like this.

Funny you should ask, I tried it with a couple words and it does appear to shift the accent making for some funny pronunciations (bass or bAss). Glad it helped! 

Very cool about the capitalization, Ben! 

Will, glad you got a solution!

Hello,

I have picked this thread from the MANY that i find on the discussion boards about Text to Speech in Articulate 360. I have been using this feature extensively of late, particularly the Matthew and Joanna voices. They have a very listenable quality to them and overall do a fine job rendering narration. I prefer them to the NeoSpeech voices provided with Captivate.

However, i notice in the threads many users struggle with pronunciations, phrasing and timing for particular text. The general workaround is to experiment with various phonetic spellings to achieve the desired sound. Yesterday, I again had to spend about an hour experimenting and found it very wasteful and frustrating—especially with deadlines looming.

I too would like to ask for some basic markup tools. Like being able to specify the verb versus noun form of a word would go a LONG way. Being able to adjust timing, speed, etc.,  would also help. The VTML aspect of NeoSpeech is very helpful.

I don't think a full VTML implementation is required but some further degree of control would be truly appreciated.

I don't know if submitting yet another enhancement request would help, there appear to be enough to give Articulate the message.  Has this actually made it to a feature design list for Articulate?

Hi Don,

While there are no current plans to implement additional editing capabilities, we are actively monitoring requests, and we appreciate you letting us know how important this is to you.

We'll let you know as we expand that feature set or make any changes, and your voice truly helps us prioritize and determine next steps. I'll be sure to share the additional insight with my team!

+1 to the desire for more control over pronunciation. Captivate definitely has Storyline when it comes to this issue. I was having an issue with the word "close" using the incorrect pronunciation. I'm also having issues with numbers not being pronounced the way I'd like. For instance, I want 245 pronounced two-forty-five instead of two-hundred-forty five. It more of an annoyance, though, as I found a workaround.

The issue I was having with the suggestions for phonetic spelling of words is that it then screwed up the closed captions. I just figured out to change the caption text after it had been created using text-to-speech. It was a small victory, but I'll take it.

I recently created an application that might address the needs listed here in this thread.  It’s a bit of a workaround though as you’ll have to go to the source that Storyline uses, Amazon Polly voices and SSML tags.  

Here’s what you do:

1. Get an Amazon Polly account (yes, there is some cost involved but doesn’t seem that prohibitive) (https://aws.amazon.com/polly/)
2. Save your scripts as separate files (MS-Word or Text)
3. Download HeroVoice TTS from the Microsoft Windows Store (fully functioning 15-day free trial)
4. Encode your files with HeroVoice TTS – apply a global setting for speed and even comma duration so your files are consistent.
5. Select the voice you want – these are the same as you’ll find in Storyline today including Neural voices (which aren’t currently available in Storyline)
6. Load each audio file into Storyline

Looking at Amazon Polly's support page for SSML tags, the 'say as' tag could be used to address the saying of numbers and the 'phoneme' tag can help with the pronunciation of words.  With that being said, I've noticed a difference between using Amazon Polly and how Storyline uses it. For whatever reason, words sometimes are pronounced differently - it might be how Storyline sends text to Amazon Polly. In HeroVoice TTS there is a preview button so you can check it before investing time to either spell it out phonetically or writing your say-as or phoneme SSML tags. :)