Repetitive transient noise removal

• US 8,326,621 B2
• Filed: 11/30/2011
• Issued: 12/04/2012
• Est. Priority Date: 02/21/2003
• Status: Expired due to Term

• Current Assignee: Malikie Innovations Limited (Key Patent Innovations Limited)
• Original Assignee: QNX Software Systems Limited (Canada) (Blackberry Limited)
• Inventors: Hetherington, Phillip A., Paranjpe, Shreyas A.
• Primary Examiner(s): He, Jialong
• Application Number:
  US13/307,615
• Publication Number:
  US 20120076315A1
• Time in Patent Office: 370 Days
• Field of Search: 704/205, 704/211, 704/226, 704/233
• US Class Current: 704/233
• CPC Class Codes: G10L 2021/02085 Periodic noise
• CPC Class Codes: G10L 21/0208 Noise filtering
• CPC Class Codes: G10L 21/0216 characterised by the method used for estimating noise
• CPC Class Codes: G10L 21/0232 Processing in the frequency domain

Abstract

A system improves the perceptual quality of a speech signal by dampening undesired repetitive transient noises. The system includes a repetitive transient noise detector adapted to detect repetitive transient noise in a received signal. The received signal may include a harmonic and a noise spectrum. The system further includes a repetitive transient noise attenuator that substantially removes or dampens repetitive transient noises from the received signal. The method of dampening the repetitive transient noises includes modeling characteristics of repetitive transient noises; detecting characteristics in the received signal that correspond to the modeled characteristics of the repetitive transient noises; and substantially removing components of the repetitive transient noises from the received signal that correspond to some or all of the modeled characteristics of the repetitive transient noises.

Claims

1. A system for attenuating repetitive transient noise, comprising:

a repetitive transient noise detector configured to determine whether an aural signal includes a repetitive transient noise based on a comparison between the aural signal and a repetitive transient noise model, where the repetitive transient noise detector comprises a processor configured to perform the comparison by fitting the repetitive transient noise model to the aural signal in a time-frequency domain, and where the repetitive transient noise detector is configured to identify the repetitive transient noise as being repetitive based on a correlation between a temporal shape of the aural signal and a temporal shape of the repetitive transient noise model, and a correlation between a spectral shape of the aural signal and a spectral shape of the repetitive transient noise model; and

a repetitive transient noise attenuator responsive to the repetitive transient noise detector and configured to attenuate the repetitive transient noise identified in the aural signal and generate a noise-reduced aural signal.

2. The system of claim 1, where the repetitive transient noise identified in the aural signal is a first repetitive transient noise, and where the repetitive transient noise detector is configured to detect a second repetitive transient noise based on a comparison between a signal and the repetitive transient noise model updated based on the one or more characteristics of the first repetitive transient noise.

3. The system of claim 1, where the repetitive transient noise detector is configured to model temporal and spectral characteristics of the repetitive transient noise identified in the aural signal.

4. The system of claim 1, where the repetitive transient noise detector is configured to update a spectral shape of the repetitive transient noise model based on spectral characteristics of the repetitive transient noise identified in the aural signal.

5. The system of claim 1, where the repetitive transient noise detector is configured to update a temporal spacing of the repetitive transient noise model based on temporal characteristics of the repetitive transient noise identified in the aural signal.

6. The system of claim 1, where the repetitive transient noise model comprises an average repetitive transient noise model created from a plurality of repetitive transient noise models.

7. The system of claim 1, where the repetitive transient noise detector is configured to update the repetitive transient noise model in response to a detection of the repetitive transient noise in an absence of speech.

8. The system of claim 1, where the repetitive transient noise detector is configured to update the repetitive transient noise model through a leaky integrator.

9. The system of claim 1, where the repetitive transient noise detector is configured to update the repetitive transient noise model based on one or more characteristics of the repetitive transient noise in response to an identification of the repetitive transient noise in the aural signal, and where the repetitive transient noise detector is configured to prevent an update to the repetitive transient noise model when a speech or speech mixed with noise segment is detected.

10. The system of claim 1, where the repetitive transient noise attenuator is constrained, in response to a detection of a vowel or another harmonic structure, to limit a transient noise correction to a value less than or equal to an average value.

11. The system of claim 1, where the repetitive transient noise detector is configured with a threshold frequency above or below which the repetitive transient noise detector evaluates signals, and where the repetitive transient noise detector is configured to update the threshold frequency over time as the repetitive transient noise model learns frequencies of repetitive transient noises.

12. The system of claim 1, where the repetitive transient noise detector is configured with a threshold frequency above or below which the repetitive transient noise detector evaluates signals, where the repetitive transient noise detector is located within a vehicle, and where the repetitive transient noise detector is configured to set the threshold frequency based on a speed of the vehicle.

13. A method of attenuating repetitive transient noise, comprising:

detecting whether a transient noise of an aural signal is repetitive based on a comparison between the aural signal and a repetitive transient noise model by fitting the repetitive transient noise model to the aural signal in a time-frequency domain;

identifying the transient noise as being repetitive based on a correlation between a temporal shape of the aural signal and a temporal shape and spectral shapes of the repetitive transient noise model, and a correlation between a spectral shape of the aural signal and a spectral shape of the repetitive transient noise model; and

attenuating the repetitive transient noise identified in the aural signal to generate a noise-reduced aural signal.

14. The method of claim 13, where the repetitive transient noise identified in the aural signal is a first repetitive transient noise, the method further comprising:

detecting a second repetitive transient noise based on a comparison between a signal and the repetitive transient noise model updated based on the one or more characteristics of the first repetitive transient noise.

15. The method of claim 13, further comprising updating a spectral shape of the repetitive transient noise model based on one or more spectral characteristics of the transient noise in response to an identification that the transient noise is repetitive.

16. The method of claim 13, further comprising updating a temporal spacing of the repetitive transient noise model based on one or more temporal characteristics of the transient noise in response to an identification that the transient noise is repetitive.

17. The method of claim 13, further comprising creating the repetitive transient noise model as an average repetitive transient noise model from a plurality of repetitive transient noise models.

18. The method of claim 13, where the step of attenuating the repetitive transient noise comprises limiting a transient noise correction to a value less than or equal to an average value in response to a detection of a vowel or another harmonic structure.

19. The method of claim 13, further comprising:

setting a threshold frequency above or below which signals are evaluated for repetitive transient noise; and

updating the threshold frequency over time as the repetitive transient noise model learns frequencies of repetitive transient noises.

20. The method of claim 13, further comprising setting a threshold frequency above or below which signals are evaluated for repetitive transient noise based on a speed of a vehicle.

21. A system for attenuating repetitive transient noise, comprising:

a repetitive transient noise detector comprising a processor configured to determine whether a transient noise of an aural signal is repetitive based on a comparison between the aural signal and a repetitive transient noise model;

where the repetitive transient noise detector is configured to perform the comparison by fitting the repetitive transient noise model to the aural signal in a time-frequency domain, and where the repetitive transient noise detector is configured to identify the transient noise as being repetitive based on a correlation between a temporal shape of the aural signal and a temporal shape of the repetitive transient noise model, and a correlation between a spectral shape of the aural signal and a spectral shape of the repetitive transient noise model;

where the repetitive transient noise detector is configured to update the repetitive transient noise model based on one or more characteristics of the transient noise in response to an identification that the transient noise is repetitive; and

a repetitive transient noise attenuator responsive to the repetitive transient noise detector and configured to generate a noise-reduced aural signal by attenuation of the transient noise identified in the aural signal as being repetitive.