2.2.4 File Conversion:
Creating Sof Files
The recognizer processes files using an AudioFile interface that accepts files in a number of formats, including an internal format, Sof, previously described in Section 2.1.6. We provide a tool, isip_make_sof, to convert external speech file formats to Sof. Below we demonstrate use of this tool to convert raw files to Sof. Other formats can be easily converted in the same manner using a modified parameter file for isip_make_sof. 
Go to the directory:
isip_make_sof speech.raw By default, this command uses a set of conversion parameters designed for 16-bit 8kHz data. See conversion parameter file to view the default parameter file for this tool. This parameter file can be configured using isip_transform_builder, a general purpose signal processing tool described in detail in Section 3. Since the configuration file is a text Sof file, users can also manually edit this file to modify some of the basic parameters of the conversion process. The above command generates a file named speech.sof that contains an Sof-formatted version of the data. By default a binary version of the audio file is created. The following command: isip_make_sof -type text -suffix _text speech.raw will generate a text version of the file. Click here to view the result of this command. Further details can be found in our on-line documentation. Note that command line arguments for all our utilities follow a uniform structure implemented using a command line parser. Three important features of this interface are (1) partial name completion (e.g., "-ty" or "-typ") for option names (e.g., "-type"), (2) help messages can be displayed at any point by simply typing "-help", and (3) options can be specified in any order. Other file formats, such as Microsoft WAV, can be converted to Sof using the same process. However, the conversion parameter file which we refer to as a recipe, must be modified. This can be done using isip_transform_builder, or simply changing the line "file_format = RAW" in the FrontEnd object in the conversion parameter file to "file_format = WAV". |
