VoiceWave Vocoder
for TMS320C3x/'C4x
by DSP Software Engineering, Inc.

Software Overview

VoiceWave is a parametric multi-rate waveform coder designed for high-quality voice, low MIPS, and 
low memory utilization. This unique, proprietary vocoder is the result of years of DSPSE's advanced 
research in vocoder technology. VoiceWave is the only high-quality vocoder that spans bit rates 
from 4.8 to 13.2 kbps, making it the most flexible vocoder available today. The algorithmic delay 
for VoiceWave is 10 ms.
DSPSE's VoiceWave encodes 8-kHz sampled speech to 13.2, 9.6, 8.0, 7.2, 6.4, or 4.8  kbps. VoiceWave 
encodes frames of 160, 16-bit linear PCM samples into 33, 24, 20, 18, 16, or 12 8-bit code words.
VoiceWave processes both speech and background signals with excellent reproduction. VoiceWave has 
numerous applications in systems that require high speech quality and low digital bit rate such as 
voice-mail systems, Internet voice passing, multimedia, voice over data, secure communications, and 
digital telephony.

Features and Benefits

•	Two full-duplex channels per 60-MHz TMS320C32 processor
•	Independent, C-callable functions
•	2-MIPS decoder
•	Compatible with real time operating systems including SPOX
•	Decoder is portable to 16- or 32-bit RISC platforms or the Pentium™
•	Rate selectable at 13.2, 9.6, 8.0, 7.2, 6.4, or 4.8 kbps
•	Designed for easy integration
•	Synchronization bit provided for frame alignment in transmissions systems

Processor and System Requirements

•	Devices supported: TMS320C3x, TMS320C4x
•	Algorithm category: Vocoder
•	Requirements:
All memory requirements are in units of 32-bit words.The MIPs ratings presented require the listed 
amounts of on-chip, zero-wait-state program and data memory, and a processing frame size of 160 
samples. All measurements were made using an executable demonstration built for a third-party PC 
plug-in board.
Function
Peak MIPS
Chip RAM
Data
Program
Encoder/Decoder 13.2 kbps
10.2/2.0
512
1.4 k/1.2 k
4.2 k/1 k
Encoder/Decoder 9.6 to 4.8
10.2/2.0
512
1.4 k/1.2 k
4.2 k/1 k
Full Duplex
12.2
512
1.7 k
5.1 k
* Significant portions of read-only program and data are shared by the encoder and decoder.

User Functions

The VoiceWave coder consists of two independent, C-callable functions that perform encoding and 
decoding operations. It also includes functions that reflect the implementation's object-based 
interface. The encoder and decoder interface with arrays of linear PCM samples and 8-bit code 
words.
VoiceWave Encoder
VWE_create (...)
Dynamically allocates and initializes data memory for encoder
VWE_delete (...)
Frees encoder's usage of dynamically allocated data memory
VWE_encode (...)	
Encodes a frame of linear PCM samples into a frame of code words
VoiceWave Decoder
VWD_create (...)
Dynamically allocates and initializes data memory for decoder
VWD_delete (...)
Frees decoder's usage of dynamically allocated data memory
VWD_decode (...)
Decodes a frame of code words into a frame of linear PCM samples

Availability

•	Source and object code currently available for licensing.

Company Background and Contact Information

DSP Software Engineering, Inc. (DSPSE), founded in 1989, is a leading provider of highly-complex 
digital-signal-processing software used in telecommunications and multimedia applications such as 
computer telephony, digital wireless, PSTN, and satellite communications, and videoconferencing. 
DSPSE sells its products to OEMs and directly to manufacturers of telecommunications products. 
Contact:	DSP Software Engineering, Inc.
		175 Middlesex Turnpike
		Bedford, MA 01730  USA
		(617) 275-3733
		Fax: (617) 275-4323
		e-mail:  info@dspse.com
		www:  http://www.dspnet.com