JournalMeSchoolFunstuffContact

Muffler User Guide

davidslife.com › School › Muffler › User Guide

This page describes the functional operation of the Muffler application.

Muffler Dialog Diagrams

This section of the user manual will list all significant parts of the two main dialogs in Muffler: (1) Muffler Main Dialog (Startup Dialog) and (2) Enhancement Parameters Dialog (Variable Configuration Dialog).

Muffler Main Dialog

The following diagram is a guide to the Muffler Main Dialog.

Muffler Main Dialog

Enhancement Parameters Dialog

The following diagram is a guide to the Muffler Enhancement Dialog.

Muffler Enhancement Dialog

Algorithm Variables

Muffler requires six main variables and two waveform audio files in order to operate correctly. In this section, the default values and ranges of each variable as well as the purpose of each waveform file will be given.

Any audio file imported into Muffler must be a one-channel (monaural), eight-bit, pulse code modulation waveform audio file. Although Muffler will accept a stereo format waveform, it will process the waveform as a monaural sound. Therefore, if one channel in a stereo waveform is NOT disabled, the Muffler basal algorithm will concatenate both waveform channels and will probably output a highly distorted signal. Secondly, the parsing code in the Muffler basal algorithm first formats a waveform into its preferred format. If the a waveform file has been subject to low-level (assembly) changes, has been downloaded from some source with a 'packet' of missing data, or subjected to a lossy compression / decompression algorithm, the Muffler algorithm may prematurely overwrite the file. The creator of Muffler is not responsible for any loss of data. Secondly, Muffler is not capable of converting sound formats. For this primary reason, it is extremely risky to input any format other than an eight bit, monaural, PCM waveform audio sound file into Muffler.

Input File - The input waveform audio file is what Muffler uses as the initial sound file that it processes. In the processing of the sound file, Muffler does NOT alter the input audio file at all. If the sound file is extremely large (consists of more than 216 bytes or longer than two minutes), the asynchronous playing function must load the whole sound file into the memory buffer or break up the sound and stream data to the buffer. In these two cases, once the sound starts, there will be a lengthy delay (approximately 30 seconds) before Muffler responds to a 'stop sound' command.

Output File - The output file is the enhanced version of the input file. An output file should be temporary, since Muffler will first clear the file in order to process the input file. In other words, any output file will be overwritten as soon as the 'Process File' button is clicked.NOTE: Since Muffler does not support real-time processing, the input file and the output file must be different.

Generalized Wiener Constant - This variable is the first entry in the list box and is named 'Gen.Wiener Cnst'. The default value for this variable is 4. The minimum value for this variable is 2 and the maximum is 10. It has one decimal place.

Spectral Floor Scaling Factor - This variable is the second entry in the list box and is named 'Spectral Floor'. The default value for this variable is 0.0005. The minimum value for this variable is 0.0005 and the maximum is 1. It has four decimal places.

Initial Noise Estimate Frames - This variable is the third entry in the list box and is named 'Init.Noise Frame'. The default value for this variable is 4. The minimum value for this variable is 2 and the maximum is 10. It has one decimal place.

Noise Adaptation Radio Box - This variable determines if noise adaptation is enabled. If noise adaptation is enabled, two variables 'Likelihood ratio' and 'Forgetting Factor' are also enabled. If noise adaptation is disabled, then these two subsidiary variables are also disabled.

Likelihood Ratio - This variable is a subvariable that pertains to the noise adaptation radio box. If the noise adaptation radio button is set to 'Yes', the likelihood ratio is the fourth entry in the list box and is named 'Likelihood Ratio'. The default value for this variable is 1.4. The minimum value for this variable is 0 and the maximum is 2. It has one decimal place.

Forgetting Factor - This variable is a subvariable that pertains to the noise adaptation radio box. If the noise adaptation radio button is set to 'Yes', the likelihood ratio is the fifth entry in the list box and is named 'Forgetting Fctr'. The default value for this variable is 1.00. The minimum value for this variable is 0 and the maximum is 1. It has two decimal places.

Tutorial

As the purpose of Muffler is to enable a user to efficiently manipulate perform speech enhancement using Ephraim's signal subspace algorithm, this tutorial will demonstrate the normal usage of Muffler.

  • Click on the 'Input File' button or use the 'Input File Edit Box' to select an existing file.
  • Click on the 'Output File' button or use the 'Output File Edit Box' to select an existing temporary file / or a new sound.
  • After you've looked at the processing variable parameters, if you need to change them, double click in the list box or click on the button labeled 'Parameters'.4
  • After re-checking the variables and the input / output files, click on the process button to initiate the algorithm.
  • After the processing step is completed, you may check the difference between both input and output sound files by selecting the appropriate choice in the sound file selection radio box and clicking the button labeled 'Play'.

Bug Report

Although Muffler is fairly robust, there are a few bugs in the application. Due to the stub interface that allows Muffler to receive run '*.exe' files as algorithms, Muffler cannot interface or communicate with its algorithm. This means that Muffler cannot (1) determine the progress of the algorithm and (2) halt the execution of the algorithm. However, this bug does not usually matter since the algorithm fully executes within a fraction of a second.

Future Enhancements

Although Muffler is fully capable as a speech enhancement application, it lacks some of the standard capabilities of standard released application. If Muffler is released in the future, it will be enabled with the following capabilities [see below].

Embedded Spectrogram Display - This capability will allow Muffler to display a spectrogram of a specified sound file.

Embedded Waveform Frequency Display - This capability will allow Muffler to display a waveform frequency of a specified sound file.

Background Thread Processing - This capability will allow Muffler to stop a stub application and not display the MS-DOS window at the same time.

Real Time Operation - This capability will allow Muffler receive sound input from an internal sound source (CD) or an external sound source (Microphone) and play or save to file an enhanced sound file simultaneously.

Multiple Format Processing / Converting - This capability will allow Muffler to process compressed sound files as well as sound in other formats (MPEG 1,2,3; AVI; VOC; CD AUDIO)