Thomas' Work Space

What does the Language-Learning-Audio-Stretcher introduced in an earlier blog post do to an audio file you feed in?

For illustration purposes, let’s have a look at a segment of a news broadcast. The example is(taken from the daily Langsam gesprochene Nachrichten by Deutsche Welle: a nice service of slowly spoken news for language learners – in my experience, however, not spoken slowly enough for North-American German students.

This timeline (X axis) shows what a computer program has automatically detected as pauses of varying length (Y axis) in the audio. Depending on a (safety) threshold which the user sets (manually, or, from experience, stored and loaded from a configuration file) in the dropdown boxes of the lower dialogue, the program attacks pauses from a certain threshold value up only:

The segment below consists of a single sentence about peace negotiations with North Korea. It is shown in the following screenshot.

• 1: transcript of the original audio file
• 1a: audio graph of the original audio file
• 2: transcript of the stretched audio file. A new line in this transcript represents a pause inserted by the software.These pauses should aid language students in review the utterance last heard in memory, and hopefully parsing it correctly.
• 2a: audio graph of the stretched audio file.
• 2b: note: non-flat audio is stretched
• 2c: note: flat lines show the pauses inserted, on top of stretching the audio.

Hearing is believing:

• segment original audio:
  dwelle-langsam-gesprochene-nachrichten-0710-ori-020120.wma
• the same segment in the stretched audio:

This software can be applied to any of numberless public domain audio books (see Project Gutenberg or Wikipedia,  audio books, as well as other free audio book sources) in mp3 or wma format (other formats can be converted). It can also can be applied to commercial audio books, if you have proper licensing.

The software comes with many options that allow you to tweak the output to your liking and needs, see prior blog post.