swirec_acoustic_adapt_suppress_adaptation

Temporarily stops self-learning activities for one or more languages.

Value

Three boolean values, where each corresponds to a different CPU load (idle, normal, busy).The third value, for busy, is also used automatically for pegged. See swirec_load_adjusted_cpu_ranges for a discussion of these ranges.

Use 1 to suppress.

DEFAULT: (depends on language)

Some languages are 0, others are 1.

Default

How to set

Shown in precedence order (see Rules of parameter precedence):

Set in the browser, using the Vendor-Specific-Parameters header in a RECOGNIZE or SET-PARAMS method.

In a parameter grammar.

In a grammar using a <meta> tag (see the text on recognition-time parameters in Parameters set in grammar files).

In the application's session xml in the <speechrecog> section of <sessionparameters>.

In Management Station set on the Nuance recognition service. If not using Management Station, set in a Recognizer configuration file (User-nrsxx.xml).

Usage

Occasionally used in various situations (see below). When used, this parameter is most common in a grammar <meta> or the application’s session.xml.

Performance

The settings correspond to CPU loads so you can disable learning during busy moments.

This parameter stops the Recognizer from collecting self-learning statistics temporarily or in specific situations. To stop all self-learning activities, use swirec_acoustic_adapt_adapt_model.

Reasons to suppress adaptation:

  • Suppress adaptation during initial deployments (to avoid adapting the test environment). In later phases of deployment, this parameter is useful to suppress certain recognition servers, languages, or advanced grammars (for example, SLMs).
  • Suppress adaptation when testing applications and grammars.
  • Use a <meta> tag in a grammar to suppress adaptation for that grammar. This is the most common use because it provides a narrow definition.

    Nuance recommends suppression for voice enrollment grammars, advanced natural language grammars, and in general any grammar that is highly unconstrained (such as a “phoneme loop” grammar).

    For example, because the per-utterance error rate is typically higher in an SLM context when compared to other grammars, it’s helpful to suppress adaptation by adding this meta to the SLM wrapper grammar.

  • Use this parameter as a throttle to reduce system load during busy periods.
  • At runtime, before a single recognition event, suppress adaptation from a VoiceXML application. This technique is seldom used because application developers usually determine when suppression is needed via the <meta> tag.

This parameter accepts values that control adaptation based on current CPU load. For most languages, the feature is 0 by default and the Recognizer automatically improves its recognition accuracy. The feature is 1 by default for some languages where the benefits of self-learning are marginal and the feature is not useful.

Precedence:

  • In MRCP the setting overrides settings in individual grammars.
  • In a grammar the setting is valid until the grammar is de-activated. VoiceXML and MRCP can override the setting for any single event.
  • In a parameter grammar the setting becomes the default for all grammars. Grammars, VoiceXML, and MRCP can override the setting.
  • In a session.xml, the setting becomes the application default.
  • When set on a recognition service in the Management Station, the setting becomes the Recognizer default for all applications using that recognition service. In this scenario, it’s best to use swirec_acoustic_adapt_adapt_model instead.

The following examples show adaptation being suppressed. The three values correspond to CPU load: idle, normal, and busy. When suppressed, the Recognizer does not adapt utterances for this language:

Browsers usually set this parameter in a RECOGNIZE request, and the parameter automatically reverts to the previous value after the recognition event. (Alternatively, browsers can use the SET-PARAMS method. See Configuring Recognizer with the browser.)

Vendor-Specific-Parameters: swirec_acoustic_adapt_suppress_adaptation=0 0 1

The example below shows the same settings in a grammar file. The setting overrides the value in any configuration file, but only for this grammar. When suppressed, the Recognizer will not adapt utterances recognized from this grammar:

<meta name="swirec_acoustic_adapt_suppress_adaptation"
content="0 0 1 "/>

For an overview of adaptation, see Controlling the self-learning recognizer.

Related parameters