Adaptive Multi-Rate Wideband

Enhanced Full Rate ( EFR ) is a specialized on speech signals, based on ACELP audio format for lossy audio data compression. Adaptive multi rate (AMR, AMR -NB ) is essentially an extension of ERA further bit rate modes. Adaptive Multi-Rate Wideband ( AMR -WB, TS 26 171, G.722.2 ) adds the original AMR essentially the ability to reproduce another octave at a reproducible frequency range added. Extended Adaptive Multi-Rate - Wideband ( AMR -WB ) is an evolution of who ACELP yet Transform Coded Excitation ( TCX ) is used in addition to offer better performance with non-voice signals can, and also stereo signals and even higher sampling frequencies possible. The Enhanced Voice Service ( EVS) of LTE Rel -10 can complement a conventional AMR -WB - layer for additional data layers that mainly an extension of the reproducible frequency range achieved but also multi-channel audio to be possible.

The company Ericsson, Nokia and VoiceAge Corporation ( Université de Sherbrooke ) hold patents on ACELP. VoiceAge collects royalties for the use of the entire patent pools.

Features and technology

The entire family of formats is essentially based on Algebraic Code Excited Linear Prediction ( ACELP ).

Adaptive Multi-Rate Wideband ( AMR-WB )

Template: Infobox file format / Maintenance / Magic number is missing template: Infobox file format / Maintenance / Developers missing template: Infobox file format / Maintenance / missing site

AMR-WB has been developed for wide-band speech connections in mobile radio systems of the second and third generation and is an international standard under the designations " TS 26171 " ( ETSI/3GPP ) and " G.722.2 " (ITU -T).

It is, with its relatively high sampling rate of 16 kHz, a so-called "wide screen" format. With the use of AMR -WB the transmitted frequency range compared with the previous standard telephone quality of 3.4 kHz to about 6.4 kHz and 7 kHz is extended by one octave. The method should also be able to transfer the mixed signals with speech and ambient noise better and allow even in noisy environments better voice quality.

There are several variants defined, each with different data rate of between 1.75 ( Silence Descriptor, SID, in silence) or 6.6 and 23.85 kbit per second. The main variation is AMR WB_12.65 12.65 kbits of data per second, which is intended for pure speech signals. In more complex signals are switched according to the high bitrate intensive variants. With bad transmission conditions can be accessed on the limited to " narrow band " play variants with 8.85 or 6.6 kbit / s. With a voice activity detection may in periods without actual useful signal transfer to so-called " Silence Descriptor" ( SID) of 1.75 kbit / s are limited (broken transmission, DTX), which contain only parameter to control a comfort noise generator.

The method works with blocks of 320 samples ( = 20 ms). With additional 5 ms lookahead results in a codec latency of 25 ms. The computational effort ( complexity) is 38.9 according to the manufacturer WMOPS.

Enhanced Voice Service ( EVS)

EVS is developed for wide-band voice calls in mobile radio systems of the third and fourth generation. A backward-compatible AMR -WB - layer, various additional data layers are set aside. Means replication of the signal from the AMR-WB - layer is synthesized an additional high-frequency signal. The harmonics of the fundamental signal for very tonal signal components generated by appropriately scaled sinusoids - in a frequency-transformed representation of the signal for different subbands of being synthesized complement signal matching sections to the selected by the output and put together in an adapted volume or - alternatively or additionally. The encoder determines the control parameters for the reconstruction process, and transmits it in an extension of the data layer. Thus, the spectral region around another octave to a super -wideband signal (up to 14 kHz) is complemented. With a further similar additional data layer of this to another octave up to cover the full range of human hearing ( "full band") can be expanded. Other possible additional streams to include parameters for reconstruction of other audio channels.

Use

AMR -WB is used according to the purpose for which it was designed in the classic third generation mobile communications. The company Ericsson has performed in Germany in the UMTS network by T -Mobile in the summer of 2006 with selected customers in the cities of Cologne and Hamburg AMR -WB mode test. Since late 2008, all Ericsson BSCs for AMR -WB are prepared in the network of T -Mobile in November 2011, it is widely available to end-users. After Telekom and Vodadafone since March 2014 also offers E- Plus to the procedures under the marketing name of HD Voice. According to the March 2014 O2 plans short to medium term is no use of AMR -WB.

Within the CAT-iq that codec for home cordless telephone according to the DECT - standard for broadband voice (16 kHz sampling rate ) is provided. It is also used for IP telephony applications. Thus, the process of various so-called softphone is used with SIP and other transmission protocols.

Software

The 3GPP has published a reference implementation for AMR -WB in the source code. VoiceAge is for Windows operating systems an experimental version for non-commercial purposes in the source code. From the Open Core project an implementation exists as free software, distributed under the terms of version 2 of the Apache license. FFmpeg enabled since March 2010, first use of the Open Core Library and developed in the summer as part of a program grant from the Google Summer of Code own free implementation of a decoder for the format.

History

The basic ACELP method was developed in 1989 at the Université de Sherbrooke in Canada. The reason for the development of AMR -WB was the search of the European Telecommunications Standards Institute (ETSI ) and the 3rd Generation Partnership Project (3GPP ) as a new wideband codec with the ability to adapt the bit rate, specify the Adaptive Multi-Rate Wideband. After a feasibility study in the spring of 1999 mid-year standardization process was initiated. The jointly developed by the company Nokia and VoiceAge method was selected after in the qualification phase in spring 2000 and the ongoing June to October selection phase in December after practice tests for the standard. After the specification was completed and approved in March 2001. Since January 2001, the process is a by the International Telecommunication Union ( ITU) recommended standard ( G.722.2 ). The method used in the 3G CDMA2000 variable - rate multimode wideband ( VMR- WB) is developed along the lines of AMR -WB and compatible with it.

For LTE ( mobile generation 3.9 / 4) from the 3GPP substantially improve the speech quality in telephone conversations is sought. The Nokia Research Center published on April 18, 2010, an elaboration of the concept.