4.7.3. Silence Compressed Record

The Silence Compressed Record (SCR) feature enables a recording with silent pauses eliminated. This results in smaller size recorded files with no loss of intelligibility. The SCR feature is enabled in the Voice.prm file which is downloaded during initialization. You must edit this file and set appropriate values for the SCR parameters for your working environment before initializing the board(s). You cannot enable this feature through the Intel Dialogic voice API.

The Voice.prm file is downloaded by default to all voice boards during the installation and configuration process. As such, SCR is available to all voice channels in the system.

The SCR parameters specify the silence threshold, the duration of silence at the end of speech before silence compression begins, the duration of a glitch in the line which does not stop silence compression, and more. Details on the parameters are provided in Section , Silence Compressed Record Parameters. After SCR is enabled in the Voice.prm file, SCR is automatically activated through use of voice record functions such as dx_rec( ).

When the audio level is at or falls below the silence threshold for a minimum duration of time, silence compressed record begins. When a short burst of noise (glitch) is detected, the compression does not end unless the glitch is longer than a specified period of time.

Silence Compressed Record Parameters

As distributed, the SCR parameters in the Voice.prm file appear as comments (each line is preceded with #). To enable the silence compressed record feature, edit this file and remove the # from the beginning of each line containing an SCR parameter. Recommended values for the SCR parameters are provided in the file.

# ================================================================
# SILENCE COMPRESSED RECORD Parameters
# To turn on SCR uncomment all of the lines in the block below.
# Recommended values are given. 
# ================================================================ 
# --- For Silence Compressed Record, uncomment the block below --- 
#PARAM 134 : 100  # SCR_T = 1 second SCR trailing silence
#PARAM 135 : 100  # SCR_PC = 100 bytes of pre-compensation
#PARAM 136 : 43   # SCR_THRES = -43dB silence threshold
#PARAM 137 : 4    # SCR_DG = 40ms of non-silence deglitch
#PARAM 138 : 1    # SCR_ON = SCR is on 
# ---- End of SCR block ----
# 

Figure 13 and Table 3 illustrate and describe the use of the SCR parameters.

Figure 13. SCR Parameters Illustrated

Table 3. SCR Parameters 

ID	Parameter Name	Description
134	SCR_T(trailing silence)	Units: 10 millisecondsRange: 0-100Default: 100 (corresponds to 1 second) The duration of silence following the end of speech before silence compression begins. This value impacts the amount of compression to be performed and thus the final file size. As you decrease the value, the amount of silence recorded between speech is decreased. You can adjust this value to suit your environment. For example, increase this value if words or sentences run together, and decrease it if the intervals of silence are too long.
135	SCR_PC(pre-compensation)	Units: bytesRange: 0 - 512Default: 512 bytesRecommended: 100 bytes The number of bytes of pre-compensation. Pre-compensation specifies the maximum length of silence that is recorded on the leading edge of speech. This prevents the beginning of speech that activates recording from being dropped (clipped) after a period of silence. Note: For best performance, use the recommended value of 100 bytes. Otherwise, the recording may become garbled.
136	SCR_THRES(silence threshold)	Units: a numeric value converted to dBRange: 20 to 50Default: 43 (corresponds to -43 dB) The audio level in the phone line below which the signal is considered noise and above which it is considered speech. When the audio level is at or below the value set in SCR_THRES for a minimum duration of time set in SCR_T, silence compression begins. The SCR_THRES numeric value is converted to a negative dB value by the firmware: 20 represents -20 dB, 21 represents -21 dB and so on up to 50 which represents -50 dB. You can adjust this value to suit a particular environment; for example, the threshold might be higher in a noisy environment. If you specify an invalid value, it is ignored and the default value is used.
137	SCR_DG(de-glitch)	Units: 10 milliseconds (ms)Range: 0 - 20Default: 4 (corresponds to 40 ms) The maximum non-silence period (glitch) that is ignored. A glitch may be a spike or short burst of noise on the line that is not speech. Silence compression continues if a glitch less than or equal in duration to SCR_DG occurs. You can increase this value if the recording includes too much noise, or decrease it if you are losing speech.
138	SCR_ON	Default: 0 Controls whether SCR is enabled. Valid values are: 0 - silence compressed record is not enabled1 - silence compressed record is enabled

Encoding Algorithms

The following encoding algorithms and sampling rates are supported in silence compressed record:

• 6 kHz and 8 kHz OKI ADPCM
• 8 kHz and 11 kHz linear PCM
• 8 kHz and 11 kHz A-law PCM
• 8 kHz and 11 kHz m-law PCM

Enabling Silence Compressed Record on Only One Board

The silence compressed record feature is enabled in the Voice.prm file. When this file is downloaded during initialization, SCR is enabled on all boards in your system.

To enable SCR on only one board in a multi-board system configuration, perform the following steps:

1. Disable the SCR parameters in the Voice.prm file.
2. Create a new parameter file that contains the SCR parameters, for example, by copying and renaming Voice.prm to Voicescr.prm, and then edit the SCR parameters in the new parameter file.
3. Download this new parameter file to the desired board by specifying it in DCM with the ParameterFile parameter. See Section 4.9, Modifying Other DCM Property Sheet Parameters for information about setting parameters in DCM.

Click here to contact Telecom Support Resources