富迪fortemedia回声消除echo cancelling白皮书white paper
需积分: 0 92 浏览量
2014-01-27
05:23:02
上传
评论 1
收藏 219KB PDF 举报
Technical White Paper on Fortemedia Voice Processing Technologies
Introduction
Fortemedia’s voice processing IC is a unique solution that enables one piece, full duplex
speakerphones to be designed into a single package previously thought impossible. Microphone
to loudspeaker separation of only 40mm is now a reality, allowing a full duplex speakerphone to
be built in a 25x40mm package (see figure 1). The purpose of this white paper is to introduce this
revolutionary voice processing technology which combines Fortemedia’s patented AS3NLF
(Adaptive Step Sized Sub-band with Non-Linear Filter) technology with AMBIN (Array
Microphone Beam-forming Integrated Noise suppression) small array microphone into a
single chip solution. Fortemedia’s single IC solution can be applied for all speakerphone
applications, and can achieve up to 65dB of acoustic echo cancellation and up 18dB noise
suppression. Fortemedia’s single IC includes ADCs, DACs, and digital interface for control,
voice, and data.
Figure 1. Comparison of Fortemedia Speakerphone and Conventional Speakerphone
Conventional Speakerphone Fortemedia Speakerphone
Background
Full duplex communication is the natural way of communicating because both parties on either
side of the communication channel can talk at the same time without having their voice cut off.
This is how people talk face to face, naturally such that either party can speak at anytime during
the conversation. Therefore, it is unnatural to communicate in half duplex mode where one side
has to wait for the other side to finish talking. In half duplex mode, if doubletalk happens, the
voice will be cut off on one side and some of the conversation will be lost.
With full duplex communication, echo, howling, and acoustic feedback become problematic if the
headset or speakerphone used cannot isolate the speaker from the microphone acoustically. The
traditional method is to separate the speaker and the microphone by keeping them far apart to
minimize acoustic leakage and to also avoid putting them in the same enclosure. However, the
size of the device cannot be smaller than the minimum distance required for the separation of the
speaker and microphone to achieve full duplex. Although DSP algorithms can be used to
overcome the issues mentioned above, most of the algorithms used cannot provide sufficient
echo cancellation to bring the speaker and microphone closer while still driving the speaker to full
volume. Most of them also cannot suppress non-linear echo to allow the speaker and
microphone to be designed in the same enclosure.
资源评论
