edu.cmu.sphinx.decoder.scorer
Class BackgroundModelNormalizer
java.lang.Object
edu.cmu.sphinx.decoder.scorer.BackgroundModelNormalizer
- All Implemented Interfaces:
- ScoreNormalizer, Configurable
public class BackgroundModelNormalizer
- extends java.lang.Object
- implements ScoreNormalizer
Normalizes a set of Tokens against the best scoring Token of a background model.
- Author:
- Holger Brandl
|
Field Summary |
static java.lang.String |
ACTIVE_LIST_PROVIDER
The active list provider used to determined the best token for normalization. |
| Methods inherited from class java.lang.Object |
equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
ACTIVE_LIST_PROVIDER
@S4Component(type=SimpleBreadthFirstSearchManager.class,
mandatory=false)
public static final java.lang.String ACTIVE_LIST_PROVIDER
- The active list provider used to determined the best token for normalization. If this reference is not defined no
nomralization will be applied.
- See Also:
- Constant Field Values
BackgroundModelNormalizer
public BackgroundModelNormalizer()
BackgroundModelNormalizer
public BackgroundModelNormalizer(SimpleBreadthFirstSearchManager activeListProvider)
- Parameters:
activeListProvider - The active list provider used to determined the best token for normalization. If this reference is not defined no
nomralization will be applied.
newProperties
public void newProperties(PropertySheet ps)
throws PropertyException
- Description copied from interface:
Configurable
- This method is called when this configurable component needs to be reconfigured.
- Specified by:
newProperties in interface Configurable
- Parameters:
ps - a property sheet holding the new data
- Throws:
PropertyException - if there is a problem with the properties.
normalize
public Scoreable normalize(java.util.List<? extends Scoreable> scoreableList,
Scoreable bestToken)
- Description copied from interface:
ScoreNormalizer
- Normalizes the scores of a set of Tokens.
- Specified by:
normalize in interface ScoreNormalizer
- Parameters:
scoreableList - The set of scores to be normalizedbestToken - The best scoring Token of the above mentioned list. Although not strictly necessary it's
included because of convenience reasons and to reduce computational overhead.
- Returns:
- The best token after the all
Tokens have been normalized. In most cases normalization won't
change the order but to keep the API open for any kind of approach it seemed reasonable to include this.