Abstract: A computer may train a single-class machine learning using normal speech recordings. The machine learning model or any other model may estimate the normal range of parameters of a physical speech production model based on the normal speech recordings. For example, the computer may use a source-filter model of speech production, where voiced speech is represented by a pulse train and unvoiced speech by a random noise and a combination of the pulse train and the random noise is passed through an auto-regressive filter that emulates the human vocal tract. The computer leverages the fact that intentional modification of human voice introduces errors to source-filter model or any other physical model of speech production. The computer may identify anomalies in the physical model to generate a voice modification score for an audio signal. The voice modification score may indicate a degree of abnormality of human voice in the audio signal.

Abstract: A computer may train a single-class machine learning using normal speech recordings. The machine learning model or any other model may estimate the normal range of parameters of a physical speech production model based on the normal speech recordings. For example, the computer may use a source-filter model of speech production, where voiced speech is represented by a pulse train and unvoiced speech by a random noise and a combination of the pulse train and the random noise is passed through an auto-regressive filter that emulates the human vocal tract. The computer leverages the fact that intentional modification of human voice introduces errors to source-filter model or any other physical model of speech production. The computer may identify anomalies in the physical model to generate a voice modification score for an audio signal. The voice modification score may indicate a degree of abnormality of human voice in the audio signal.

Abstract: A computer may train a single-class machine learning using normal speech recordings. The machine learning model or any other model may estimate the normal range of parameters of a physical speech production model based on the normal speech recordings. For example, the computer may use a source-filter model of speech production, where voiced speech is represented by a pulse train and unvoiced speech by a random noise and a combination of the pulse train and the random noise is passed through an auto-regressive filter that emulates the human vocal tract. The computer leverages the fact that intentional modification of human voice introduces errors to source-filter model or any other physical model of speech production. The computer may identify anomalies in the physical model to generate a voice modification score for an audio signal. The voice modification score may indicate a degree of abnormality of human voice in the audio signal.

Abstract: Provided are methods and systems for calibrating a distributed sensor (e.g., microphone) array using time-of-flight (TOF) measurements for a plurality of spatially distributed acoustic events at the sensors. The calibration includes localization and gain equalization of the sensors. Accurate measurements of TOFs are obtained from spatially distributed acoustic events using a controlled signal emitted at known intervals by a moving acoustic source. A portable user device capable of playing out audio is used to produce a plurality of acoustic events (e.g., sound clicks) at known intervals of time and at different, but arbitrary locations based on the device being moved around in space by a user while producing the acoustic events. As such, the times of the acoustic event generation are known, and are spatially diverse. The calibration signals emitted by the acoustic source are designed to provide robustness to noise and reverberation.

Abstract: Provided are methods and systems for finding the location of sensors (e.g., microphones) with unknown internal delays based on a set of events (e.g., acoustic events) with unknown event time. A localization algorithm may iteratively run to compute the acoustic event times, the observation delays, and the relative locations of the events and the sensors.

Abstract: Provided are methods and systems for calibrating a distributed sensor (e.g., microphone) array using time-of-flight (TOF) measurements for a plurality of spatially distributed acoustic events at the sensors. The calibration includes localization and gain equalization of the sensors. Accurate measurements of TOFs are obtained from spatially distributed acoustic events using a controlled signal emitted at known intervals by a moving acoustic source. A portable user device capable of playing out audio is used to produce a plurality of acoustic events (e.g., sound clicks) at known intervals of time and at different, but arbitrary locations based on the device being moved around in space by a user while producing the acoustic events. As such, the times of the acoustic event generation are known, and are spatially diverse. The calibration signals emitted by the acoustic source are designed to provide robustness to noise and reverberation.

Abstract: Provided are methods and systems for finding the location of sensors (e.g., microphones) with unknown internal delays based on a set of events (e.g., acoustic events) with unknown event time. A localization algorithm may iteratively run to compute the acoustic event times, the observation delays, and the relative locations of the events and the sensors.