CODEWITHSUNDEEP
iosobjective-cavspeechsynthesizer
Greyeye
Greyeye

Reputation: 109

AVSpeechSynthesizer with ios8

HI did anyone tried AvSpeechSynthesizer on iOS 8 ? I did quick app on Xcode 6 but no audio comes out, I ran the same on on the Xcode 5 and works without a hitch.

sample code from http://nshipster.com/avspeechsynthesizer/

NSString *string = @"Hello, World!";
AVSpeechUtterance *utterance = [[AVSpeechUtterance alloc] initWithString:string];
utterance.voice = [AVSpeechSynthesisVoice voiceWithLanguage:@"en-US"];

AVSpeechSynthesizer *speechSynthesizer = [[AVSpeechSynthesizer alloc] init];
[speechSynthesizer speakUtterance:utterance];

Bug with Xcode6?

== Edit === Looks like bug/issue with iOS 8 sim, iOS7.1 Sim with Xcode6 works fine..

Upvotes: 5

Views: 6865

Answers (5)

Matt
Matt

Reputation: 2359

This is still an issue on 8.0.2 (Xcode 6.0.1, testing on iPad mini 1).

This is only an issue if you are setting the voice property for AVSpeechUtterance to a language other than the default dialect (in my case English (U.S.) is default). To be clear, the issue can be resolved by forcing AVSpeechSynthesizer to speak an empty AVSpeechUtterance instance in the non default language first. That last part is important.

Below is a simple getter that implements a fix.

- (AVSpeechUtterance *)utterance {
    if (!_utterance) {
        ...
        _utterance = [AVSpeechUtterance speechUtteranceWithString:@"hello world"];
        [_utterance setRate:0.2f];
        [_utterance setVoice:[AVSpeechSynthesisVoice voiceWithLanguage:language]];

        // iOS 8 bug fix
        AVSpeechUtterance *dummyUtterance = [AVSpeechUtterance speechUtteranceWithString:@" "];
        [dummyUtterance setVoice:[AVSpeechSynthesisVoice voiceWithLanguage:language]];
        [self.synthesizer speakUtterance:dummyUtterance];
    }
    return _utterance;
}

Upvotes: 4

starea
starea

Reputation: 341

It indeed a bug in XCode 6 and iOS 8.0.2. As long as you configured the "voiceWithLanguage" property, it won't speak the sentence for the first time. So the current solution is to speak an empty string initially with the desired language.

Upvotes: 1

Ali Beadle
Ali Beadle

Reputation: 4526

EDIT: the comment about recreating the synthesizer below appears to be relevant to iOS8.0.0 and GM only - in iOS8.0.2 it does not seem to be necessary and user848555's workaround works fine for me now. Interestingly user848555's workaround DOES seem to still be necessary (at least if you change languages), despite comments I have seen on other threads that it has been fixed in iOS8.0.2.

So this seems to be a bug related to low quality voices - it is referenced in the release notes here (search for Speech Synthesis on this page):

[https://developer.apple.com/library/ios/releasenotes/General/RN-iOSSDK-8.0/][1]

The offical Apple workaround is to download a high quality voice for the voice that is not working (by default one voice is always high quality - US English in English speaking nations, presumably this changes in other nations) which is why it seems to work if you stick to the default voice.

user848555's workaround does not work in isolation if (like me) you are constantly changing voices. However if you recreate the speechSynthesizer as well on each uttereance that appears to work. Here is my code (it is specific to my app so you will need to change a few things but it should illustrate the point). It uses user848555's fix of a blank utterence plus a recreate on each utterence. 

-(void)speakGerman:(NSString*)text{
    if (floor(NSFoundationVersionNumber) <= NSFoundationVersionNumber_iOS_6_1 ||
        [self.dataController.speakSetting integerValue] == 2) return;

    // iOS8 fix
    speechSynthesizer = [[AVSpeechSynthesizer alloc]init];
    AVSpeechUtterance *bugWorkaroundUtterance = [AVSpeechUtterance speechUtteranceWithString:@" "];
    bugWorkaroundUtterance.rate = AVSpeechUtteranceMaximumSpeechRate;
    [speechSynthesizer speakUtterance:bugWorkaroundUtterance];

    AVSpeechUtterance *utterance;
    
    utterance = [AVSpeechUtterance speechUtteranceWithString:text];
    utterance.voice = germanVoice;
    utterance.rate = 0.2;
    utterance.preUtteranceDelay = 0.0;

    [speechSynthesizer speakUtterance:utterance];
}

Ali [1]: https://developer.apple.com/library/ios/releasenotes/General/RN-iOSSDK-8.0/

Upvotes: 1

ICL1901
ICL1901

Reputation: 7778

I found this class solved the problem (or at least is a good workaround) for me...

https://github.com/dale-moore/iOSTextToSpeech

Upvotes: 0

user848555
user848555

Reputation: 181

Yup, its a bug. As of iOS8 GM, it seems the first attempt to get AVSpeechSynthesizer to speak an AVSpeechUtterance will result in silence.

My workaround is to make AVSpeechSynthesizer speak a single-character utterance immediately after it is initialized. This will be silent, and afterwards your AVSpeechSynthesizer will work as normal. 

AVSpeechUtterance *bugWorkaroundUtterance = [AVSpeechUtterance speechUtteranceWithString:@" "];
bugWorkaroundUtterance.rate = AVSpeechUtteranceMaximumSpeechRate;
[self.speechSynthesizer speakUtterance:bugWorkaroundUtterance];

Upvotes: 9

Related Questions