• 0

Question

Hey gang,

I have a media application (Bedia, free available here: http://www.blissgig.com/display.aspx?id=13 ) that I have added speech recognition. The problem is that on some lower end systems the SR causes the audio playback to become "off" from the video playback, usually by only a fraction of a second. I am using WMP as the audio/video playback.

The question is, I am thinking of having a threaded class that does the monitoring for the speech recognition, but before I go down that path I thought I'd ask for any thoughts? Is there anything I should be aware of or......

I know the question is a bit vague, I am attempting to avoid mentioning specifics so that any and all issues for this process are brought up.

Thank for your time.

9 answers to this question

Recommended Posts

  • 0

If I understand well, you're currently handing the audio data to a speech recognition API before handing it to WMP, which can cause the audio to become out of sync? It's pretty strange that this is happening, but I'm not familiar with WMP.

In any case, doing this on separate thread would work as long as you're actually handing out the data to the processes (WMP and SR) in parallel and not having one waiting on the other, otherwise you're not gaining anything by using multiple threads.

  • 0

Thanks Dr.

The speech commands affect Play, Pause, other playback controls as well as menu commands. I really don't know why the audio becomes unsynced so the idea of moving the SR to another thread is a shot in the dark. I have completed the code in a test bed and will move it into the app in the next few days. I will drop a note here to let everyone know how it goes.

The nice part of the SR being on another thread is that some of the processing (is a command valid, etc) is within that thread and SHOULD not affect any playback. I just return a string that informs the base app what I need to do, just as if the user had pressed a mouse button.

Have a great day.

James

  • 0

Thanks Dr.

The speech commands affect Play, Pause, other playback controls as well as menu commands. I really don't know why the audio becomes unsynced so the idea of moving the SR to another thread is a shot in the dark. I have completed the code in a test bed and will move it into the app in the next few days. I will drop a note here to let everyone know how it goes.

The nice part of the SR being on another thread is that some of the processing (is a command valid, etc) is within that thread and SHOULD not affect any playback. I just return a string that informs the base app what I need to do, just as if the user had pressed a mouse button.

Have a great day.

James

Did you do it delegate style? Or are you constantly checking a string for a value?

  • 0

Did you do it delegate style? Or are you constantly checking a string for a value?

Yes I am using a Delegate, thanks for asking. Only when acceptable conditions have been met is a string sent back to the calling thread. This is why I thought about going this route, all the work to insure that this is an acceptable command before affecting the app. I really do appreciate you asking to make sure.

Peace,

James

  • 0

Yes I am using a Delegate, thanks for asking. Only when acceptable conditions have been met is a string sent back to the calling thread. This is why I thought about going this route, all the work to insure that this is an acceptable command before affecting the app. I really do appreciate you asking to make sure.

Peace,

James

I know some people will thread things but then sit there waiting for the thread to finish/do something instead of carrying on and doing what needs to be done away from the thread and acknowledging when the thread has something for it. That's the reason I asked.

But yea, logically if the thread is just sitting in a loop like:

wait for command

process command

if command is good raise the event

repeat

and the main program is sitting in a

play movie/sit in idle state

**Event raised**

check event

do event

repeat

And there is no reliance on one or the other to preform individual tasks and you aren't sitting there in a:

while (x != y) or while(myThread.ThreadState == ThreadState.Running)

Then I don't see why it wouldn't work, and it should work better than a non-threaded system.

  • 0

Thanks again Firey,

Your diagram statement is exactly what is going to happen (I haven't added my new class to my app yet. I almost always create a test bed application to insure that it is working on it's own before adding it to the real application)

I still find it strange that the audio becomes unsynced.... but when I shut down the SR option, which disposes of the SR components, the issue goes away. I state this in case anyone else ever has the issue that they know that someone else has seen it.

Have a fun day.

  • 0

Thanks again Firey,

Your diagram statement is exactly what is going to happen (I haven't added my new class to my app yet. I almost always create a test bed application to insure that it is working on it's own before adding it to the real application)

I still find it strange that the audio becomes unsynced.... but when I shut down the SR option, which disposes of the SR components, the issue goes away. I state this in case anyone else ever has the issue that they know that someone else has seen it.

Have a fun day.

One thing to check, would be CPU usage. I know that in some cases High CPU usage can cause things to become un-synced. If the playback has full video decoding frames can be ready and display but the CPU can't keep up with keeping things (like the Audio) synced properly.

  • 0

One thing to check, would be CPU usage. I know that in some cases High CPU usage can cause things to become un-synced. If the playback has full video decoding frames can be ready and display but the CPU can't keep up with keeping things (like the Audio) synced properly.

Interesting thought. The app has run for two+ years on this lightwight machine but with the addition of SR it could be hitting high. I will check this out, thanks.

This topic is now closed to further replies.
  • Posts

    • Formatting would be awesome if fully customised and then roll out the settings as a company wide standard.
    • I just want to know why the volume control feature in the browser version keeps appearing then disappearing, here today - gone tomorrow. Now it's gone again. Annoying as hell.
    • When you have only lies and blood libels to present, you've lost. And how hard is it for you to look up BBC and AP? You hate Jews, just admit it.
    • This article does not make sense and it’s very evident, the bug wasn’t confirmed to exist, because the application itself runs on editions of Windows which make it impossible to run Hyper-V VMs on, “Basically, Claude Desktop on Windows spins up a 1.8GB Hyper-V virtual machine if you use Claude Cowork or agent mode even once.”, suggests the bug is caused by something else otherwise it would only happen to users with the capability of running Hyper-V VMs and that’s only if Hyper-V was enabled which is by default not enabled. The true issue is that the author who reported the bug used the incorrect terminology, it’s actually a container, not a VM that is being started by the way. There were users on macOS reporting the same issue, Hyper-V, doesn’t exist on that platform obviously. Every single user who confirmed the problem in the last week was either using macOS making this issue impossible to be relevant to them, or was a comment talking about WSL, which the author indicated 3 months ago was disabled.
    • “To prevent hallucination issues and keep the database clean, the platform uses a multi-agent verification loop to check code quality.” - This absolutely isn’t happening. ”Before the massive rise of LLMs, which tanked its traffic by about 50% over the last couple of years, Stack Overflow was the go-to website for millions of programmers seeking coding solutions.” - It still is trusted, because the community of users, blocked other users from submitting false and misleading answers generated by AI agents that can be convinced something it generated is actually incorrect even if it’s actually correct. The company itself doesn’t know what it wants to be so it keeps trying and failing to introduce AI features to the platform. This effort will also fail, already submitted one successful poison pill, look forward to generating more.
  • Recent Achievements

    • One Month Later
      Sopa flores earned a badge
      One Month Later
    • First Post
      StaticMatrix earned a badge
      First Post
    • Week One Done
      StaticMatrix earned a badge
      Week One Done
    • Rookie
      lamborghiniv10 went up a rank
      Rookie
    • One Month Later
      pinnclepd earned a badge
      One Month Later
  • Popular Contributors

    1. 1
      +primortal
      524
    2. 2
      PsYcHoKiLLa
      211
    3. 3
      +Edouard
      160
    4. 4
      Steven P.
      98
    5. 5
      ATLien_0
      83
  • Tell a friend

    Love Neowin? Tell a friend!