Algotron: TI DSP Software For Telecoms  
   A TI DSP Third Party Network Member
     

    Home | Modem Data Pumps | Telephony | Audio | Utilities | Consultancy | Contact Us

 

Audio Summary |G.722 Speech Coder |G.726 Speech Coder |G.729 speech coder

P.56 Speech Level Meter Software for Texas TMS320C54x/C55x & C2xx/C5x DSPs

This software implements the ITU P.56 Speech Level Meter function (Method B). The P.56 function is intended primarily for speech level measurement. Over a time interval, the algorithm obtains the active level of an irregular signal such as speech by discerning and measuring in only the active signal periods. In parallel, the long-term level is measured as the true RMS value of the signal overall, irrespective of active or inactive signal periods. A further result given by the process is the activity factor, that is the proportion of the active signal period to the total measurement time.

The software is in the form of three callable subroutines: one is for initialisation where a user-set measurement time in samples is input, one is for processing of each new sample, and one is for computing the results. If required, the results routine can be called at any time prior to the expiry of the set measurement time. Otherwise, the results are calculated automatically at the end of the set time. Further calls of the main per-sample routine cause the process to be repeated. The sample rate can be 8kHz, 9.6kHz, 16kHz, or 32kHz.

The user is responsible for sample input and for accepting the three results: long-term level, active level, and activity factor. The software can be operated in a multi-channel mode: a pointer is set to each per-channel block of memory at each sub-routine call. By making multiple sub-routine calls, the code can be operated on blocks of samples. The code is load-balanced so there is no MIPS variation per sample, except when the results are computed.


Performance: (*accuracies of activity-related results are subject to the P.56 measurement process)

  • Measurement time range: up to 74 hours (8ks/s) in sample periods

  • Dynamic range: 68dB

  • Resolution/accuracy*: 0.1dB in level measurement, 0.1% in activity factor

DSP Load:

  • MIPS: 0.77( 8.0kHz s.r.)

  • Data memory: 2(general)+58(per channel) words

  • Program memory: 0.46k words


Availability: NOW, with implementer's guide - integration support offered

Audio Summary |G.722 Speech Coder |G.726 Speech Coder |G.729 speech coder