<object data="test.wav" type="audio/wav" width=100 height=50>
alternative text
</object>
An audio clip. User agent may present some kind of embedded graphical controls for controlling the playback. If this is the case, the controls should be scaled to the size of 100*50 pixels.
The spec doesn't say implicitly what should happen if the width and height attributes are present, but the user agent has no visual rendering for audio OBJECTs. However, since width and height override the natural object size, a 100*50 blank space might be the most compliant rendering in a case like this.
User agents which cannot play embedded WAV sounds should display the words "alternative text".