Auto-CS doc

Module CuedSpeech.whenhand.transition

Class WhenTransitionPredictorDuchnowski1998

Description

Predict hand transition moments with (Duchnowski et al., 1998) method.

Paul Duchnowski, Louis D. Braida, David S. Lum, Matthew G. Sexton, Jean C. Krause, Smriti Banthia AUTOMATIC GENERATION OF CUED SPEECH FOR THE DEAF: STATUS AND OUTLOOK https://www.isca-speech.org/archive/pdfs/avsp1998/duchnowski98avsp.pdf

In section 2.2 of this paper: "We found that cues are often formed before the corresponding sound is produced. To approximate this effect we adjusted the start times of cues to begin 100 ms before the boundary determined from acoustic data by the cue recognizer."

Reference is:

Paul Duchnowski, Louis Braida, Maroula Bratakos, David Lum, Matthew Sexton, Jean Krause A SPEECHREADING AID BASED ON PHONETIC ASR https://isca-speech.org/archivev0/archivepapers/icslp1998/i980589.pdf

In section 3.2 of this paper: "We observed that human cuers often begin to form a cue before producing the corresponding audible sound. To approximate this effect we adjusted the start times of the cues to begin 100 ms before the boundary determined by the cue recognizer."

In this class, no transition time values are estimated. The returned intervals are [A1-0.1,A1-0.1] and [A1-0.1,A1-0.1].

Constructor

Instantiate (Duchnowski et al. 1998) hand transition moment's predictor.

View Source

def __init__(self):
    """Instantiate (Duchnowski et al. 1998) hand transition moment's predictor.

    """
    super(WhenTransitionPredictorDuchnowski1998, self).__init__()
    self._description = MSG_DESCRIPTION_DUCHNOWSKI

Public functions

predict_position

Predict [M1,M2] the moments when the hand is moving.

Returns

(m1: float, m2: float)

View Source

def predict_position(self, **kwargs) -> tuple:
    """Predict [M1,M2] the moments when the hand is moving.

        :return: (m1: float, m2: float)

        """
    m1 = max(0.0, self.a1 - 0.1)
    m2 = max(0.0, self.a1 - 0.1)
    return (m1, m2)

predict_shape

Predict [D1,D2] the moments when fingers are changing.

Returns

(d1: float, d2: float)

View Source

def predict_shape(self, **kwargs) -> tuple:
    """Predict [D1,D2] the moments when fingers are changing.

        :return: (d1: float, d2: float)

        """
    d1 = max(0.0, self.a1 - 0.1)
    d2 = max(0.0, self.a1 - 0.1)
    return (d1, d2)