Problem solve Get help with specific problems with your technologies, process and projects.

Cat got your tongue?

Cat got your tongue?
Tom Lancaster

One common quality issue with VoIP is clipped speech. Usually, this happens at the beginning of a sentence or after a pause in a conversation, where a half-second or so of conversation is mysteriously dropped. If you encounter this issue in your VoIP network frequently, the culprit may actually be a feature called Voice Activity Detection (VAD).

VAD is good because it allows you to conserve quite a bit of bandwidth by transmitting packets only when someone is talking. During periods of silence, no bandwidth is used. However, there are a number of problems. The first is that it takes some time for most networking gear to realize someone is talking and switch from suppressing to transmitting. Your equipment is going to make this transition as fast as it possibly can. If that's not fast enough for you, there isn't much you can do to speed it up, but there are a few tips to help you avoid the problem in the first place.

First, lengthen the amount of silence required before suppression is activated. This course of action is appropriate when you get clipping between phrases in a sentence, or even between words. Remember that silence suppression is mostly useful because only one person talks at a time, so there's no need for the other party to be transmitting while they are only listening. If silence suppression is activating while a person is talking, you need to increase this timer. (As an example, in Cisco's IOS, this is done with the VOICE VAD-TIME <milliseconds> command.) Since speech characteristics vary widely by culture and location, there isn't a single optimal value for this timer.

Second, if you have clipping when conversation is low, such as whispering, then try adjusting the noise threshold. Remember that most implementations of VoIP define a level of noise so that anything below that value is considered background noise and isn't transmitted, while anything above that value is considered speech and is transmitted. When people speak softly, the hardware may hear them speaking, but think it's background noise and ignore it. (As an example, in Cisco's IOS, this threshold can be adjusted via the oddly named MUSIC-THRESHOLD <-db> command.)

If neither of these adjustments is sufficient to alleviate your clipping problems, consider changing CODECs. A few of the popular CODECs offer versions with and without VAD.

Thomas Alexander Lancaster IV is a consultant and author with over ten years experience in the networking industry, focused on Internet infrastructure.

Did you like this tip? Why not let us know? Send an email and sound off.

Related Book

Cisco Voice over Frame Relay, ATM, and IP
Author : Steve McQuerry
Publisher : Cisco Press
Published : Apr 2001
Summary :
Cisco Voice over Frame Relay, ATM, and IP is a direct complement to the Cisco authorized training course of the same name. Based on the content of the CVoice course, this book provides an intermediate-level treatment of Cisco voice technologies. The overall objective of the book is to teach engineers how to design, integrate, and configure voice over Frame Relay, ATM, and IP of enterprise or managed network services using various Cisco 2600, 3600, 3810, and 5300 multiservice access devices.

This was last published in May 2001

Dig Deeper on IP Networking

Start the conversation

Send me notifications when other members comment.

By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

Please create a username to comment.