This implementation of a G.165 Line Echo Canceller (LEC) is designed to reduce or even cancel electrical echoes in the transmission of speech signals. The tail length of the echo suppression can be adjusted to system requirements. The algorithm is required in DECT telephony systems and similar applications.

Features

* Software controlled Coefficient Adaptation
* Software controlled Leakage
* Software controlled Non-linear Processor (NLP)
* Software controlled Tone Disabler
* Double Talk Detector with Adaptation Control
* Programmable allowable Tail Length

Specifications

* ~ 10 MIPS per channel (32 msec tail)
* < 2K bytes program memory
* 76 + 32*T bytes data memory (T = tail in msec)

Platforms

The algorithm is available for Analog Devices Blackfin and Texas Instruments TMS320C54x/C55x. The API for all platforms is designed to be independent of signal source and sink. The code is fully re-entrant allowing multi-channel applications to be easily realized.
We have demo code available for standard evaluation platform, e.g. C55 DSK or Blackfin EZKit which we can provide to our (potential) customer on short notice. If required, we can port the existing code to other platforms - all you need to do is ask.

Datasheets

Our implementation of a G.167 Acoustic Echo Canceller is designed to cancel acoustic feedback between a loudspeaker and a microphone in loudspeaking audio systems. The tail length of the echo suppression can be adjusted to system requirements. The algorithm is required in speaker phones, teleconferencing devices, voice-control systems or alike.

Features

• Software controlled Coefficient Adaptation
• Software controlled Leakage
• Software controlled Non-linear Processor (NLP)
• Software controlled Howling Control
• Double Talk Detector with Adaptation Control
• Programmable Tail Length

Specification

• ~10 MIPS per channel (32 msec tail
• < 2K words program memory
• max. 250 msec tail @ ~50 MIPS
• 80 + 32*T bytes data memory (T = tail in msec)
• Single Talk Attenuation > 45db
• Double Talk Attenuation > 30db
• Convergence ~ 20dB/sec
• ITU G.167 compliant

The algorithm is currently available for Analog Devices Blackfin and Texas Instruments TMS320C54x/C55x, implementations for other platforms on request. All implementations allow multi-channel applications due to the re-entrant and flexible structure of the algorithm.

Demo versions are available on request for evaluation platforms such as EZKit or DSK

Datasheets

G.167 AEC for Analog Devices Blackfin
G.167 AEC for Texas Instruments C5000
G.167 AEC for Texas Instruments C6000

If you wish to receive further information please, contact us.

Adaptive-Multirate Wide-Band Speech Codec (ACELP)

The G722.2 recommendation describes the detailed mapping from input blocks of 320 speech samples in 16-bit uniform PCM format to encoded blocks of 132, 177, 253, 285, 317, 365, 397, 461, and 477 bits and vice versa. The sampling rate is 16 kHz leading to a bit rate for the encoded bit stream of 6.60, 8.85, 12.65, 14.25, 15.85, 18.25, 19.85, 23.05, or 23.85 kbit/s. The coding scheme for the multi-rate coding modes is Algebraic Code Excited Linear Prediction Coder (ACELP). The multi-rate wideband ACELP coder is referred to as AMR-WB. The codec described in this recommendation also utilizes an integrated Voice Activity Detector (VAD).

Our implementation of G.722.2 is available for Blackfin platforms and can be demonstrated on BF533-EZLite or simulated on PC platforms.

Datasheet for Blackfin

If you wish to receive further information please, contact us.

G.729A/B 8.0 kbps Voice Codec

G.729AB is a reduced complexity version of G.729 speech coder standard from the ITU-T, for compressing the toll quality speech (8000 samples/second) at 8kbps. Annex B implements optional silence-compression techniques to reduce the transmitted bit rate during the silent periods of speech (voice activity detection). Typical applications of this speech coder are in telephony over packet networks, like Voice-over-Internet-Protocol (VoIP).

We currently offer implementations of the G.729A/B Codec for the following DSP families: Texas Instruments TMS320C55x, TMS320C54x, TMS320C62x, TMS320C64x and Analog Devices Blackfin.

Datasheet for TI C54x/C55x
Datasheet for TI C62x/C64x
Datasheet for Blackfin

If you wish to receive further information please, contact us.

Wideband Speech-Codec (7kHz Bandwidth)

Datasheet TI C54x/C55x
DatasheetTI C62x/C64x
Datasheet Blackfin

For more information, please contact us.

G723.1 Voice Codec 5.3/6.3 kbps

G.723.1 is a dual rate speech coder standard from the ITU-T, for compressing the toll quality speech (8000 samples/second). Typical applications of this speech coder are in telephony over packet networks, like Voice-over-Internet-Protocol (VoIP). This speech coder is also used for coding the speech component in video conferencing applications and is part of the H.324 family of standards.
This codec supports two bit rates, 5.3 and 6.3 kbps. Both bit rates share the same short-term analysis techniques for processing the speech. For long-term analysis of speech, the algorithms used are different. For 5.3 Kbps coder, Algebraic Code Excited Linear Prediction (ACELP) principles are used where as in 6.3 kbps coder, Multi Pulse-Maximum Likelihood Quantization (MP-MLQ) techniques are used.

We currently offer implementations of the G.723.1 Codec for the following DSP families:
Texas Instruments TMS320C55x, TMS320C54x, TMS320C62x, TMS320C64x and Analog Devices Blackfin.

Datasheet for TI C54x/C55x
Datasheet for TI C62x/C64x
Datasheet for Blackfin

For more information, please contact us.