Overview

Audio processing utility is a pre-processing application used for manipulating the input audio (tones, and voice) files with impairments such as delay, noise, acoustic echo, line echo, and double talk. An echo file is generated if a corresponding line or acoustic echo filter is specified. It analyzes the given coefficients files and calculates the applied delay & ERL. The degraded audio output file can then be visually analyzed or used for further speech analysis with GL's Voice Quality Testing, Digital Echo Canceller & other applications.

APU (Audio Processing Utility) is designed to produce a specific configurable degraded file for a given input file. An echo file can also be generated if a corresponding line or acoustic echo filter is specified. In addition, if two files are specified as input files, then, the echo path characteristics will be calculated.

---

Typical applications

• P.56 active speech level algorithm testing
For a well-defined input file, APU will generate a precisely controlled degraded file. This degraded file can be used to test P.56 active speech level algorithm using applications


• VQT (Voice Quality Testing)
The input voice file and the controlled degraded file can be fed into the VQT algorithm and test VQT applications


• EC (Echo Canceller) testing
The input (reference file) file and the echo file can be used to test the EC. APU can generate line echo and acoustic echo files, and test line EC and acoustic EC, which makes it unique.


• Offline echo analysis
The offline echo analysis function of APU will “analyze the stimulus” & degraded signal files and calculate the echo path filter coefficients, ERL and delay.

---

Main Features

• Generates echo files with appropriate acoustic & hybrid echo
• Generates degraded voice files with noise, power level, & interference
• Supports high quality voice over IP, with substantial packet delay and loss
• Generated files can be used to test VQT, ECs, & P.56 active speech level
• Supports A-law encoded PCM, µ-law encoded PCM, Windows PCM (*.wav), and 16-bit Intel linear PCM audio files
• Supported Codecs are G.711 µ-law, G.711 A-law, G.723.1, G.726, G.729, G.729a, G.729b, G.729ab, G.729e, G.729g, iLBC (20ms and 30ms frame)
• Analyzes filter coefficients, ERL (Echo Return Loss), and delay from the generated files
• Supports line echo filter model of ITU-T G168 specification and three proprietary acoustic echo filter models
• Graphical representation of output files

---

Operations

The degraded files can be generated either by adding Noise, Filter, and/or Double Talk. Commonly, only the noise file is added to the input file, to generate a degraded (A_in) file. The figure depicts the APU algorithm where a noise file is added to the reference (input) file, to generate a degraded (A_in) file. The reference file and the degraded file can be used for voice quality testing.

These degraded files may be further filtered as per line/acoustic echo models to generate echoed files. Double talk files can also be added to the echo signals. The degraded and echoed files can be used in ECs (echo cancellers).

Accordingly the main operations of APU are:

• Add a noise file to input (tone / voice file). Noise can either be white noise or any noise files of the type *.ala, *.ula or *.pcm
• Adjust power level for the Noise file
• Generates A_in file (A-law, µ-Law, Wave, or 16-bit Intel linear PCM file formats)
• (Optional) Process A_in based on echo path models (line or acoustic echo) to generate B_in file
• (Optional) Further add noise or double talk signals to the generated B_in file

Thus, two files A_in and B_in are generated in this process. A_in consists of input and noise with proper power level. B_in is the echo version of A_in file along with noise and/or double talk (optional).

Adding Noise

White Noise of specified power (dBm0) and SNR (dB) or Noise from specified file is added to the input file to generate degraded files (A_in).

Adding Filter

The generated noise (A_in) files are filtered as per line or acoustic echo models to generate echo (B_in) files. For "Line Echo", users can choose M1 ~ M8 as per ITU-T G.168 specifications (Version 08/2004) options or "Others" for user-defined filters. A1 ~ A3 (acoustic filters) are GL proprietary implementations of the acoustic filters such as small office, medium office and large office. The application also provides "No Filter" option for not generating B_in files. Normally, one can use APU to generate controlled degraded files for VQT applications.

Adding Double Talk

Double talk files can also be added to the echo version of degraded file along with noise to generate echoed degraded file. Double talk files can also be added to the echo version of A_in file along with noise to generate B_in files.

"None" option helps to generate pure echo files. Users can either select tones of required frequency (Hz) and power level (dBm0) or choose a double talk file of certain power level to generate degraded B_in files. The power of the selected file can also be adjusted to the user defined level.

Input and Output Files

Input to APU can be tone of specified frequency (Hz) and power level (dBm0) or could be voice from a specified file. Supported input files are A-Law, µ-Law, or 16 bit linear pcm.

APU supports four output file formats - A-law ("*.ala"), µ-law ("*.ula"), wave (".wav"), and linear PCM ("*.pcm").

The output file length can be restricted either to the same length as that of input file or up to specified time in seconds until the specified file length is reached. In the latter process, if the specified output file length is shorter than the original input file, then, the former part of the input file is used and the rest of the input signal is cut off.

If the specified output file length is longer than the input file, then, the input file is concatenated and processed till required output file length is reached.

One may specify the gap in the concatenation process, which refers to idle code (silence) in milliseconds. The "No Gap" option wraps the input file.

The out put files generated A_in and B_in can be viewed along with the average power values in dBm0.

Offline Analysis

With this feature the APU generated files can be analyzed offline.

A_in is the stimulus signal, and B_in is the signal with echoes. The file format of both these files should be same. APU supports A-law ("*.ala"), µ-law ("*.ula"), 16-bit linear ("*.pcm"), and Windows PCM wave format (".wav")

Along with the Ain and Bin files, the filter order (number of filter coefficients) should be set to the 'delay + dispersion' value. Ideally, to capture the characteristics of the filters, the filter order should be greater than or equal to the hybrid model length (significant coefficient). The echo path filter coefficient will be generated in *.txt file name, while the ERL (Echo Return Loss) and delay parameters are displayed in the window. The delay is defined as the maximum filter coefficient position and the ERL algorithm is calculated as follows:

---

Resources

Please Note: The XX in the Item No. refers to the hardware platform, listed at the bottom of the Buyer's Guide, which the software will be running on. Therefore, XX can either be ETA or EEA (Octal/Quad Boards), PTA or PEA (tProbe Units), XUT or XUE (Dual PCIe Express) depending upon the hardware.

* Specifications are subject to change without notice.