LAB 17: Use of Landmarks and Features for Speech Recognition

Labs: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17

The main purpose of this laboratory is to introduce you to analyzing and annotating acoustic events (landmarks and features) in two sentences. You can choose various ways to analyze the sentences.

Here are the resources that you will need:

Lab 17 Handout (PDF)

Utterances

In this lab, you will use the recorded utterances stored in the lab17 directory. The utterances are labeled senta, sentb, sentc, and sentd.

If you do not have access to the lab17 directory, you can download example recordings of the utterances needed for this lab.

The Tools

You can use either one of the following two analysis tools:

  • xwaves - used for spectrum and waveform analysis (UNIX®/Linux®). If you use this tool, you also have the option of using the transcriber script. This script configures xwaves for the types of tasks done in this lab.
  • xkl - used for spectrum and waveform analysis (UNIX®/Linux®)

Reference Readings

Stevens, K. N. "Toward a model for lexical access based on acoustic landmarks and distinctive features." J Acoust Soc Am 111 (2002): 1872-1891.