3.4 Sound source separation fails


Read this section when a sound that cannot be considered speech is output or heavy distortion occurs, when intending to separate a mixture of sounds from a microphone input.


When separating with the GHDSS module, perform the separation in accordance with the following conditions.

Confirm that the source localization has been successfully completed

Since the GHDSS module uses source localization results (number of sound sources) for inputs, the separation performance degrades if localization failed. Modify the network files so that the localization results are displayed and confirm whether the system localizes sounds successfully. For further information, see the recipe, Sound source localization fails.

Confirm that GHDSS separates sounds properly

To determine whether the GHDSS has been performed properly, save a result of GHDSS using SaveWavePCM and confirm that separation was successful. If separation failed in GHDSS , see Parameter tuning of sound source separation

Confirm that post-filtering was performed properly

When PostFilter or HRLE are inserted just after separation processing, separation may fail due to improper parameters. Save the prost-filtered sound and confirm that the post-filtering was successful. If the post-filtering failed, see bsec:Separation: 006]Reducing the leak noise by post processing



See Also

Sound source localization fails
Parameter tuning of sound source separation
Reducing the leak noise by post processing