XMOS VocalFusion XVF3510 User Manual page 55

Table of Contents

Advertisement

The delay estimation process supports two stages:
}
Fine delay estimation, during which AEC adaption is paused. Fine delay estimation searches
over a limited delay window to detect small changes in delay. If the delay correction is not
resolved the coarse estimation is triggered.
}
Coarse delay estimation, re-purposes the AEC to detect larger delays. During estimation, the
AEC does not perform cancellation. Once the delay is detected, and delay correction made the
AEC restarts and converges based on the delayed signals.
Possible causes that may trigger coarse estimation:
}
Device reset (if initial delay estimation is enabled, default: enabled)
}
Host changing applications
}
Large volume changes between the reference and the loudspeaker play-back
}
User equipment changes, such as switching from TV audio output to playing the audio through
a soundbar
Possible causes that may trigger fine estimation:
}
Host toggling between audio devices (such as using a voice assistant application and listening
to music at the same time) This is typically only seen in USB configurations
}
Host processor performance leading to poor USB buffer management
The characteristics and specification of the ADEC function is shown below:
Table 4-15 ADEC characteristics
NAME
VALUE
Maximum delay
± 150ms
correction
Coarse
With
good
estimation time
2-5 seconds
<1 second.
Fine estimation
If fine-grain estimation fails,
time
coarse-grain
triggered.
The configuration commands are shown below:
Table 4-16 Automatic Delay Estimator parameters
COMMAND
GET_DELAY_SAMPLES
SET_DELAY_SAMPLES
XM-014232-PC
DESCRIPTION
The maximum delay that can be added to either the microphone
channel or the reference channel.
reference
SNR:
During this time AEC is disabled. Note that estimation will not start
unless reference is available and loudspeakers are playing back.
During this time adaptation in the AEC is disabled. Note that
estimation will not start unless reference is available and loudspeakers
estimation
is
are playing back.
TYPE
VALUE
uint32
[0 .. 2399]
DESCRIPTION
Change the number of samples of input delay
at the sample rate 16kHz.
The delay is applied to either the
reference or the microphone in
put according to the delay direction.
This provides a maximum delay of +/- 150mS.
NOTE
S
A
55

Advertisement

Table of Contents
loading

This manual is also suitable for:

Vocalfusion xvf3510-intVocalfusion xvf3510-ua

Table of Contents