armHigh Density Conferencing


Narrowband, Wideband, and Mixed Narrowband/
Wideband capable

Adaptive Digital provides support throughout all project phases; customers have access to our technical team directly over the phone or via email.

High Density IP Voice Conferencing Software

High Density (HD) IP Voice Conferencing with Mixed Narrowband and Wideband Channels.


  • Narrowband, Wideband, and Mixed Narrowband/Wideband capable
  • Number of conference participants is user configurable
  • Participants can be added and deleted from conference at any time
  • Automatic Level Control
  • Overflow Protection
  • Voice Activity Detection
  • Noise Suppression
  • Dominant Speaker Selection
  • Multi-conference capable
  • Functions are “C” callable
  • Variable Frame Size
    Build that can support different frame sizes among conference members
  • Preemptive Member Assignment
    When a conference member is assigned preemptive status, all other members’ signals will be suppressed and the preemptive member’s signal will be the only signal included the conference sum.
  • Priority Member Assignment
    Normally, conference is configured with a number of dominant speakers, typically three. This means that regardless of how many conference members are present, only the three loudest members will be summed at any given time. By assigning priority status to a conference member, that member’s signal will always be added to the sum.


ADT HD Conferencing WBNB / WB / NB is available on the following Platforms: Other configurations are available upon request.

Platform High performance High efficiency Ultra-high efficiency
ARMv8-A Cortex-A72 / A73 Cortex-A53 Cortex-A32 / A35
ARMv7-A Cortex-A15 / A17 Cortex-A8 / A9
ARMv7-M Cortex-M7 Cortex-M3 / M4 / M7 Cortex-M23
Retired/Legacy - ARM9E & ARM11

The conference algorithm is available in a narrowband version (nb), which operates at the typical telephony sampling rate of 8 kHz, as well as a wideband version, which operates at an audio sampling rate of 16 kHz. The wideband version (wb) is suitable to be used in high-end conferencing equipment as well as in VoIP applications in which wideband audio is supported. Also available is the mixed rate version. That is able to bridge together both narrowband and wideband conference (wbnb) parties into a single conference.


The Adaptive Digital Technologies Conferencing software is designed to provide conference call capability to telephone systems as well as to voice and video conference servers. Our conferencing algorithm adds the active conference input signals together to form a composite signal. Before sending the composite signal back to each conference party, that party’s transmission is removed from the composite signal to avoid the perception of echo.

You might ask, what’s the big deal? Why can’t I simply add the signals together?

Adding the signals together works well enough if the number of conference participants is very small and there is no need to support mixed narrowband and wideband simultaneously. Of course, you do need to subtract each party’s own speech from the conference composite (sum) as mentioned above to avoid the perception of echo.

As the number of conference participants increases, we run into a few more issues to deal with. For example, each participant presumably has some level of background noise. The noise level may be low as is the case in an office environment (without any fans running.) The noise level can also be quite high – as is the case for a person on a cell phone while driving. And it can be anywhere in between. If we were to add all the input signals blindly, the noise would accumulate more and more as the number of conference participants increased. Furthermore, when using fixed point arithmetic, the summation of many signals – some of which include speech signals – can cause overflow or clipping, a very undesirable condition.

Our algorithm uses a variety of techniques to combat these issues. For example, only a few “dominant” speakers’ signals are added to the conference at any given time. This reduces the number of signals being added. Furthermore, noise suppression is employed on all input channels. So even when there is significant background noise that would otherwise bleed into the conference sum, the noise suppressor reduces the extent of such noise. Automatic Level Control is employed to combat overflow and clipping as well as to compensate for different amounts of network loss that is seen in party’s input signal. We have integrated all these algorithms and optimized them to work together to achieve optimum voice quality in high-density conference situations.

Finally, Adaptive Digital’s conference algorithm includes support for mixing narrowband (8 kHz sampled) and wideband (16 kHz sampled) participants into the same conference. While this is not brain-surgery in and of itself, combining this feature with the other features in a way that preserves voice quality is quite important.

Not to be overlooked is the fact that we have combined all this functionality using a minimum of processor resources (MIPS and Memory). After all, if you need high density conferencing, you probably want to do so with a minimum number of processors and as little memory as possible.

All that said, in addition to the more sophisticated, voice quality enhancing features, we added a simple but useful feature. Sometimes it is desirable to allow one (or more) participants to take precedence to others in the conference summation regardless of voice activity. We call these participants “priority” participants. The user can set conference participants to be priority participants and set them back to “normal participants” at any time during an active conference.


ARM Devices


All Memory usage is given in units of byte.

Function Conf Type #8k Members #16k Members MIPS Program Memory Per Conference Data Memory Per Member Data Memory Scratch Memory

Mixed 8k Variable FS 8 0 6.9 18.1k 128 356 4004
Mixed 8k Variable FS 4 4 34.7
Mixed 8k Variable FS 0 8 59.8
Mixed 16k Variable FS 8 0 57.4
Mixed 16k Variable FS 4 4 33.9
Mixed 16k Variable FS 0 8 13.4


Cortex - A8 / A9 / A15

All Memory usage is given in units of byte.
Function Conf Type #8k Members #16k Members MIPS Program Memory Per Conference Data Memory Per Member Data Memory Scratch Memory

Mixed 8k Variable FS 8 0 6.6 17.3k 128 356 4004
Mixed 8k Variable FS 4 4 33.3
Mixed 8k Variable FS 0 8 57.9
Mixed 16k Variable FS 8 0 56.5
Mixed 16k Variable FS 4 4 32.6
Mixed 16k Variable FS 0 8 10.9

FS = Frame Size

WBNB = mixed Wide Band, Narrow Band

MIPS characterized using a 8-channel conference.
Mixed 8K/16K conference characterized with half of the members at 8 kHz and half of the members at 16  kHz members

We specify MIPS (Millions of Instructions Per Second) as MCPS (Millions of Instruction Cycles Per Second). Unless otherwise specified, peak MIPS are indicated.


Conf_ADT_init (…) Initialize a conference

Conf_ADT_run (…) Perform conferencing function

CONF_ADT_addMember(…) Add a member to a conference

CONF_ADT_removeMember(…) Remove a member from a conference

CONF_ADT_setPriority(…) Set a conference member as a priority speaker

CONF_ADT_clearPriority(…) Clear a conference member’s priority status

CONF_ADT_close(…) Close a conference