Siemens Microelectronics, Inc., introduces the Acoustic Echo Canceler (ACE), a powerful DSP-based chip that provides a full-duplex speakerphone in analog and digital phones.

The ACE (PSB 2170) does not require a learning tone and adapts quickly to changes in the acoustic environment, such as people moving around or a door opening. Suitable for telephone and video conferencing, the device can be used in mobile phones, analog feature phones, ISDN phones and PBX applications.

The ACE joins a series of voice-processing ICs from Siemens, including the recently introduced SAM, a programmable digital answering machine combined with speakerphone. The SAM and ACE are pin compatible and software compatible making multiple designs very easy.

The on-board DSP provides for DTMF tone generation and detection, flexible ringing detection, programmable side gain, transducer correction filters, call progress tone detector, caller ID decoder, general purpose parallel port up to 16 bits, and independent gain for all channels.

The ACE features two algorithms -- n Full-band: -20 dB ERLE @ 60 ms, <1 ms delay, full band n Sub-band: -30 dB ERLE @ > 50% ms, 43 ms delay.

The full-band algorithm introduces negligible delay and is better suited for real-time applications such as conference room speakerphones. The sub-band algorithm offers additional echo suppression enhancement and is better suited for video applications where delay is introduced for lip synching.

This is by no means the first DSP-based IC for full-duplex speakerphone applications but it is certainly the most complete that I have yet seen. The fact that it is pin-compatible with the SAM part will make it attractive to the companies who are trying to offer the most from a single design layout and the resultant economy of parts with one PCB. The tricks needed to cope with different room environments and reflectivity are now fairly well understood and it would be surprising if Siemens have not heeded those lessons and done it properly. With a DSP on-board echo detection and cancellation is also extraordinarily simple. The part is fabricated in a 3-V CMOS process and supports IOM-2 compatible channels and can offer a dedicated interface to Siemens dual-codec, PSB 4851.

I do worry about the statement about "delay is introduced for lip synching (sic)" being interpreted by a user as an assumption that delay is introduced in the audio path to compensate for the lip-sync delay caused by the video processing part of, for example, a teleconference; that is not so, with any delays in this part being small while digital-video processing delays will be highly variable depending on the encoding, transport and decoding process.

The dual algorithms that are provided are extremely nice, further increasing the market attractiveness for purchasing volume. That, of course, is only valid if the resulting cost of the part does not reflect the addition of too many "hooks" that many designs might not use. That does not seem to be the case here where the starting price of the PSB 2170 ACE is in accordance with the other parts that are on offer, at $6.50 for 100,000 piece lots. The part is in an 80-pin QFP package.