Only return iOS voices that work #24
Conversation
Only one voice per locale works in iOS (versions 12 & 13 and maybe earlier versions ). This change brings back the ios.js file and some of its former code to detect iOS. It deletes the iOS8 & iOS9 voices from ios.js in the lib folder (was that erroneously left over?) It then filters the voices returned by iOS and only returns the ones that will work.
lib/speak-tts.js
Outdated
| @@ -36,6 +36,7 @@ function () { | |||
|
|
|||
| var listeners = (0, _utils.isNil)(conf.listeners) ? {} : conf.listeners; | |||
| var splitSentences = (0, _utils.isNil)(conf.splitSentences) ? true : conf.splitSentences; | |||
| var splitPhrases = (0, _utils.isNil)(conf.splitPhrases) ? true : conf.splitPhrases; | |||
There was a problem hiding this comment.
what is this option for ? How does it compare to splitSentences ?
There was a problem hiding this comment.
Sorry, looks like I messed up working with this PR and included my internal changes. Not sure how useful this will be to the general public, but I'm trying to sync a moving picture to the TTS and I need to know when there is talking and when there is silence. splitPhrases splits on other punctuation characters that cause a pause in the TTS (like the comma).
There was a problem hiding this comment.
alright, i'd be grateful if you could remove it. Even better, splitSentences could probably be refactored so that it can serve both purposes (by passing a list of splitting chars for instance) ?
There was a problem hiding this comment.
A custom list of splitting chars would work for me and keep me from having to use a separate branch. I'll work on that.
Happy to remove this change, but I'm just not too familiar working with PRs...that's how I ended up with this in here in the first place :)
There was a problem hiding this comment.
I think I 'll do a massive refactor of this lib in the next month anyway, I'll include your changes. In the meantime you can use your own version and close this PR, or update the PR and I'll merge it.
It's up to you, let me know !
There was a problem hiding this comment.
ok, removed the splitPhrases option. I'll work on making splitSentences take a custom list of splitting chars
src/speak-tts.js
Outdated
| @@ -166,8 +188,8 @@ class SpeakTTS { | |||
| if(this.pitch) utterance.pitch = this.pitch //0 to 2 | |||
| utterance.text = sentence | |||
|
|
|||
| // Attach event listeners | |||
| Object.keys(listeners).forEach(listener => { | |||
| // Attach event listeners, and make sure onerror and onend are included | |||
There was a problem hiding this comment.
onerror & onend are used by speaktts to resolve/reject the speak promise, and this should happen even if the user did not pass customized versions of those liseteners into the init function.
There was a problem hiding this comment.
if no customized listener is passed, this is going to attach undefined variables ?
There was a problem hiding this comment.
No, newListener is a function that calls the user-provided listener (in this scenerio that is NULL and it won't be called due to "fn && fn(data)" ) but it also checks if the listener is for "onend" or "onerror" in order to resolve/reject the speak promise.
Prior to this change, if no custom listener for onend and onerror is passed, then the speak promise will not resolve or reject.
There was a problem hiding this comment.
I see, thanks for the explanation. I think it could be resolved by giving a noop function as a defaut value for onEnd et onError
| return false | ||
| } | ||
| export const filterIOSVoices = (voices) => { | ||
| let selectableVoices = [ |
src/utils.js
Outdated
| @@ -5,6 +5,16 @@ export const splitSentences = (text = '') => text.replace(/\.+/g,'.|') | |||
| .map(sentence => trim(sentence)) | |||
| .filter(Boolean) | |||
|
|
|||
| export const splitPhrases = (text = '') => text.replace(/\.+/g,'.|') | |||
splitSentences can now be true, false, or set to a string of custom characters to deliminate the sentences. An error is thrown if [a-zA-Z] are passed (a user sending in 'true' could be in for a world of hurt figuring that out.) Any number of consecutive splitting characters can be used, and the sentence will only be split once.
Only one voice per locale works in iOS (versions 12 & 13 and maybe
earlier versions ). This change brings back the ios.js file and some of
its former code to detect iOS. It deletes the iOS8 & iOS9 voices
from ios.js in the lib folder (was that erroneously left over?) It then
filters the voices returned by iOS and only returns the ones that will
work.