VMR-WB
— Source-controlled Variable Bit Rate
Wideband Compression
The Variable Rate Multi-Mode Wideband
(VMR-WB) codec is the mandatory speech
codec for cdma2000® wideband telephony
and multimedia streaming services. Designed
by Nokia and VoiceAge, VMR-WB is based
on the AMR-WB/G.722.2 codec. VMR-WB operates
in the range from 0.8 to 13.3 kbps, including
one mode of operation (mode 3) that is
interoperable with AMR-WB/G.722.2 (at
12.65 kbps and below).
|
|
|
Quality
This table compares the quality of VMR-WB with
source control at several average bit rates
for a speech signal that is 60% active speech
and 40% background noise/silence.
VMR-WB
mode |
VMR-WB at average
bit rate |
Quality |
G.722 at
fixed bit rate |
1 |
5.5 kbps |
= |
56 kbps |
2 |
4.5 kbps |
> |
48 kbps |
3* |
6.7 kbps |
= |
56 kbps |
* Interoperable
with AMR-WB/G.722.2.
Back to top
Technical Background
VMR-WB is a source-controlled variable bit rate
(VBR) codec. The VBR speech coding concept is
crucial for optimal operation of CDMA systems
(such as cdmaOne and cdma2000). In source-controlled VBR coding, the
codec operates at several bit rates, and a rate
selection mechanism is used to determine the
bit rate suitable for encoding each speech frame
based on the characteristics of the speech signal
(e.g., voiced, unvoiced, transient, background
noise). The goal is to attain the best speech
quality at a given average data rate (ADR).
The codec can operate in different modes by
tuning its rate selection algorithm to obtain
different ADRs. Codec performance is improved
by increasing the ADR. The system imposes the
mode depending on the network capacity and the
desired quality of service. This gives the codec
a mechanism of trade-off between speech quality
and system capacity.
Rate Sets
CDMA systems use rate sets comprised of 4 bit
rates referred to as full-rate (FR), half-rate
(HR), quarter-rate (QR), and eighth-rate (ER).
CdmaOne and cdma2000 systems support two rate
sets, referred to as Rate-Set I and Rate-Set
II. The following table shows the source-coding
bit rates and corresponding channel bit rates
inclusive of error protection bits for variable-rate
codecs with a rate selection mechanism operating
in Rate-Set I and Rate-Set II.
Operating rates of a variable-rate codec with
rate selection mechanism operating in Rate-Set
I and Rate-Set II
|
FR |
HR |
QR |
ER |
Rate-Set I |
|
|
|
|
Source-coding bit rate
(kbps) |
8.55 |
4.0 |
2.0 |
0.8 |
Channel bit rate including
error detection bits (kbps) |
9.6 |
4.8 |
2.4 |
1.2 |
Rate-Set II |
|
|
|
|
Source-coding bit rate
(kbps) |
13.3
|
6.2
|
2.7 |
1.0
|
Channel
bit rate including error detection bits
(kbps) |
14.4 |
7.2 |
3.6 |
1.8 |
Back to top
Operating
Modes
VMR-WB has four operating modes in Rate-Set
II and one in Rate-Set I.
Modes 0 to 3 operate in Rate-Set II. Modes 0, 1, and 2
are specific to CDMA systems (i.e., TIA/EIA/IS-95,
cdma2000) with mode 0 providing the highest
quality and mode 2 the lowest ABR. Mode 3 is
interoperable with AMR WB/G.722.2 at 12.65 kbps
and below. Its ADR is slightly higher than the
ABR of mode 0.
Mode 4 is compliant
with cdma2000 Rate-Set I with an ADR slightly
lower than mode 2 and with quality equivalent
to or better than mode 2.
VMR-WB
operating modes
VMR-WB
mode |
Description |
Approximate
speech average data rate (kbps)
|
Bit rate for 40% speech
activity (kbps) |
0 |
“Premium” |
12.8 |
5.7 |
1 |
“Standard” |
10.5 |
4.8 |
2 |
“Economy” |
8.1 |
3.8 |
3 |
Interoperable with
AMR-WB/G.722.2
|
13.3 |
6.1 |
4 |
Rate-Set I mode |
~ 8.1 |
~ 3.8 |
The system selects
one of the operational modes depending on the
traffic conditions. The transition from one
mode to another is seamless and memoryless (i.e.,
there is no transition period to achieve the
ADR of the new mode), and there is no need to
transmit the mode information to the decoder.
The system can also force maximum and minimum
rates.
Back to top
Narrowband
Operation
The narrowband codecs developed in CDMA for
Rate-Set I operation consist of QCELP8 (IS-95),
EVRC (IS-127), and SMV. In Rate-Set II operation,
QCELP13 (IS-733) is used.
While the design
of VMR-WB focused on wideband input and output
signals, VMR-WB accepts narrowband input signals
sampled at 8 kHz, and it can synthesize narrowband
speech. The codec lookahead varies depending
on the sampling frequency of the input and the
output, being 13.75 ms for wideband input and
output and 15.0625 ms for narrowband input and
output.
Technical
Highlights
Back to top
VMR-WB Standard for
cdma2000 Telephony
Approved as a standard by 3GPP2 in July 2004,
the VMR-WB codec was also adopted by the 3GPP2
as the mandatory speech codec for cdma2000® wideband telephony services. The VMR-WB wideband
speech codec is defined, described and recommended
in these standards:
Back to top
RTP Packet Format Definition
This specification defines the Real-Time Protocol
(RTP) payload format for packetizing VMR-WB
encoded audio signals into the RTP. The RTP
payload format definition enables use of the
codec in RTP packet-switched networks in applications
like streaming, and it enables interoperability
with existing codec transport formats on non-IP
networks.
IETF (2006) RFC 4348, “Real-Time
Transport Protocol (RTP) Payload Format for
the Variable-Rate Multimode Wideband (VMR-WB)
Audio Codec,” specifies an RTP payload
format to be used for the VMR-WB speech codec.
The payload format is designed to be able to
interoperate with existing VMR-WB transport
formats on non-IP networks. A media type registration
for VMR-WB RTP payload format is included. This
specification also shows how VMR-WB-enabled
terminals can interoperate directly with AMR-WB
terminals without transcoding. http://www.ietf.org/rfc/rfc4348.txt?number=4348
Technical background information for this article
came from a draft article on wideband codec
standardization targeted for publication in
IEEE Communications in May 2006.
Back to top