Tasker and the "say" Action (Google TTS)

[Justin Eltoft]

I would like to make a recommendation to the developers of tasker.  See this reddit where I posted recently.  I would like to suggest that Tasker do the following:

1. check if the "say" action is going to use a high quality (~250MB) TTS engine

2. if it is, then check if the engine is in RAM already

3. if it is NOT, then instead use the low/standard TTS engine for this "say" action and load the high quality when the overall task is done 

I've measured it, and if the TTS engine is not in RAM, standard takes about 1 second to load and play TTS the first time.  But high quality takes 8-10 seconds!!!  So any audio ducking that is done is actually totally done at the wrong time for the first time "say" speech with high quality.  

Is this a good place to post this, or is there a better way to make this suggestion for new work to developers?

Justin

[Matt R]

As a workaround, you can keep the TTS engine in memory (mostly) if you uncheck "reduce resource usage" in the Tasker preferences.

Matt

[Pent]

As a workaround, you can keep the TTS engine in memory (mostly) if you uncheck "reduce resource usage" in the Tasker preferences.

... and do a very short Say on zero volume triggered by a Monitor Start event to ensure that it's always ready.

There are some difficulties with what you suggest. Toughest is 'check if the engine is in RAM already'. It might be e.g.
mid-load (which you can't check) and other related issues.

Pent

[GJD]

Try to use the IVONA TTS from the play store. I found this one faster react and it has a more natural voices.

https://play.google.com/store/apps/details?id=com.ivona.tts

Op woensdag 14 mei 2014 17:20:34 UTC+2 schreef Justin Eltoft:

[RudeboyX]

I get the slow loading of google TTS (HQ) which I don't mind as long as it still runs the action, but what really caught me out the other day was a "failed to initiase TTS" error from tasker. I thought this issue was resolved by extending the time out on the say action? seems I still get it.

To be honest there isn't a lot of difference between the standard and HQ TTS engine other than the hugh file size, it sounds almost the same so think I will have to revert to the standard quality voice, as this seems like a never ending problem. several years now at least........ (started with Ivona TTS)