一个语音识别基于应用程序，我的工作非常适用于所有版本的Andr​​oid从API 8（Android 2.2的）。启动

A speech-recognition based app I am working on works well on all versions of Android starting from API 8 (Android 2.2).

但在一个的Nexus S 4G（安卓4.1.1），RecognitionListener将简单的暂停约1分钟，然后发出通过其onError()回调。

But on a Nexus S 4G (Android 4.1.1), RecognitionListener will simply halt for about 1 minute, then issue an ERROR_SERVER via its onError() callback.

如果讲的是onReadyForSpeech哔哔），它会表现得正常预期。

If spoken to within 1-2 seconds (of that onReadyForSpeech bleep), it will behave properly as expected.

在什么改变软糖可以解释这种现象？

What changed in JellyBean that could explain this behavior?

更重要的是，有没有办法让它表现得像在旧版本的Andr​​oid？ （即守听，然后发出ERROR_SPEECH_TIMEOUT如果默认内没有说过话的10秒）

More importantly, is there a way to make it behave like in the older versions of Android? (i.e. keep listening, then issue ERROR_SPEECH_TIMEOUT if not spoken to within the default 10 seconds)

推荐答案

我最终使用由 @scriptocalypse

感谢您@scriptocalypse！

Thank you @scriptocalypse!

（我还没有机会尝试通过 @HoanNguyen 提出的解决方案，因为它实现它是一个服务超出了我的小应用程序的范围，但它看起来很有希望。谢谢@HoanNguyen了。）

(I haven't had the chance to try the solution suggested by @HoanNguyen because it implements a service which is beyond the scope of my tiny app, but it looks promising. Thank you @HoanNguyen, too.)