Share via

ActivationSignalDetectionTrainingDataFormat Enum

  • Reference

Definition

Namespace:
Windows.ApplicationModel.ConversationalAgent
Edit

Specifies the activation signal training data formats supported by the ActivationSignalDetector for the digital assistant.

public enum class ActivationSignalDetectionTrainingDataFormat
/// [Windows.Foundation.Metadata.ContractVersion(Windows.Foundation.UniversalApiContract, 655360)]
enum class ActivationSignalDetectionTrainingDataFormat
[Windows.Foundation.Metadata.ContractVersion(typeof(Windows.Foundation.UniversalApiContract), 655360)]
public enum ActivationSignalDetectionTrainingDataFormat
var value = Windows.ApplicationModel.ConversationalAgent.ActivationSignalDetectionTrainingDataFormat.voice8kHz8BitMono
Public Enum ActivationSignalDetectionTrainingDataFormat
Inheritance
Enum
ActivationSignalDetectionTrainingDataFormat
Attributes
ContractVersionAttribute

Windows requirements

Device family
Windows 10, version 2004 (introduced in 10.0.19041.0)
API contract
Windows.Foundation.UniversalApiContract (introduced in v10.0)

Fields

Audio44kHz16BitMono 6

Training data is generic audio in 16-bit 44kHz mono.
Audio44kHz8BitMono 5

Training data is generic audio in 8-bit 44kHz mono.
Audio48kHz16BitMono 8

Training data is generic audio in 16-bit 48kHz mono.
Audio48kHz8BitMono 7

Training data is generic audio in 8-bit 48kHz mono.
AudioOEMDefined 9

Training data is generic audio in a format specified by a hardware provider.
OtherOEMDefined 10

Training data is in a format specified by a hardware provider.
Voice16kHz16BitMono 3

Training data is voice audio in 16-bit 16kHz mono.
Voice16kHz8BitMono 2

Training data is voice audio in 8-bit 16kHz mono.
Voice8kHz16BitMono 1

Training data is voice audio in 16-bit 8kHz mono.
Voice8kHz8BitMono 0

Training data is voice audio in 8-bit 8kHz mono.
VoiceOEMDefined 4

Training data is voice audio is defined by an OEM.

Remarks

Digital assistant applications can train keyword detectors to more accurately recognize an individual user's voice by applying algorithmic customizations to the detector based on speech data (the detector provides these customizations). For example, training a spoken keyword detector to only detect the keyword when spoken by a specific person.

This is achieved through a series of ActivationSignalDetectionConfiguration training steps, where each step consumes a logical fragment of speech input data.

Applies to

See also