I recently listened to an episode of the Rezzed and Confused SL podcast, produced by Itazura Radio. Among the several always-interesting topics discussed on the show was a great idea to improve the voice audio quality.
Itazura’s podcast usually provides commentary on the intermittent official Linden Lab blog postings, after a reading by his abused and truly scary assistant “Audrey”. On this episode Itazura discussed the difficulties of setting up voice, in reaction to the blog post on using SL as a conferencing tool. If you haven't heard this 'cast, I strongly recommend it. Audrey is a little hard to take, but bear with her to the end and you'll have a good laugh.
Voice, we love it or we hate it. I happen to love it, and use it by default nowadays. Of course, there are moments when I must revert to text mode, typically because the majority of the crowd is texting and not voicing.
Why do I like voice? Two reasons, I guess. One, it’s simply vastly faster to communicate, particularly complex information such as negotiating a deal with a client. Two, I don’t do role-play, where a voice that doesn’t match the role could cause catastrophic and hilarious results. I suppose there’s a third reason too: I’m not a guy playing a female avatar.
But as I use voice with many people, the same problems keep coming up:
Itazura’s podcast usually provides commentary on the intermittent official Linden Lab blog postings, after a reading by his abused and truly scary assistant “Audrey”. On this episode Itazura discussed the difficulties of setting up voice, in reaction to the blog post on using SL as a conferencing tool. If you haven't heard this 'cast, I strongly recommend it. Audrey is a little hard to take, but bear with her to the end and you'll have a good laugh.
Voice, we love it or we hate it. I happen to love it, and use it by default nowadays. Of course, there are moments when I must revert to text mode, typically because the majority of the crowd is texting and not voicing.
Why do I like voice? Two reasons, I guess. One, it’s simply vastly faster to communicate, particularly complex information such as negotiating a deal with a client. Two, I don’t do role-play, where a voice that doesn’t match the role could cause catastrophic and hilarious results. I suppose there’s a third reason too: I’m not a guy playing a female avatar.
But as I use voice with many people, the same problems keep coming up:
- Volume is too low, and you can’t hear them.
- Volume is too high, and you must crank them down
- Can’t connect - no “white dot”
- They’re too far away and thus too faint to hear
And my number one problem that simply drives me totally fricking insane: Feedback from the other end let’s me hear myself talking. This is so blindingly awful and horrendously embarrasing I want to slay myself each time it happens. It's as if someone just clipped 70 points off of your I.Q. and you suddenly became a dumbass who doesn't know how to talk.
We’ve all been there; you begin speaking, and a second or two later you hear your voice. Then, curiously, your voice stops abruptly for some reason. Then you realize it was because you stopped speaking in mid-sentence to listen to yourself.
OK, it’s not that bad. Just ignore it. Try again. You resume speaking, and it works for a while but then you
Augh. It happened again.
The feedback happens, of course, due to two main causes:
- Your correspondent does not have a headset or earbuds and you are hearing yourself come out of their speakers
- Your correspondent has a headset, but it is improperly placed on their head, and with volume sufficiently loud, your voice leaks out from behind their ears and goes back into their microphone
Horrifying, I’m sure you’d agree.
But how do we solve this? Itazura has a wonderful idea, as I mentioned earlier. The idea is to include a “call me to test” feature, similar to what is done on Skype. The voicer can “speak”, and then hear themselves exactly as others would hear them. They’d realize soon enough how badly they have set up their microphone, and fix it.
I’m thinking there’s another approach to fixing the feedback issue. And that might be a much more sophisticated audio processing back-end that watches for identical (or similar) waveforms being sent back and filters them out automatically.
Until either of those fixes takes place, we’re just going to have to be patient, and keep saying, “Can you hear me” a whole lot.
virtual business, building virtual products and exploring the virtual world.
3 comments:
I find voice too hard to follow when there are several people talking at the same time. It's confusing. And if the sound quality is bad, it's very disruptive to the conversation to have to ask the person to keep repeating themselves.
I also found that the absence of visual cues made it difficult to tell whether the other person has finished talking, or is just pausing before continuing. We have awkward silences or accidentally start talking at the same time. It gets better the more you get to know the person coz then you get to know their speaking style and the way they think/communicate, but in the beginning it's awkward.
Moreover, if it were something like negotiating a deal, I'd like to have it in text so I have a log that I can refer to later... then the client couldn't claim I'd promised something I didn't!
Skype manages to silence the feedback pretty well when not using a headset, but SL fails miserably.
I wonder if there's a JIRA we can vote up for implementing an equivalent of the skype call testing bot.
One thing i've seen quite commonly that you didn't mention is people with improperly set soundcards; they got their soundcards set to record the mix of the microphone and wave-out (and everything else); basicly when a software tries to record audio it records everything that is being sent to the speakers (regardless of what the speakers are actually sending out, they might even be unplugged but you still hear your echoes).
Post a Comment