Long term monitoring of transmission and voice activity patterns for regulating gain control

Patent metadata

Field	Value
Publication number	US-9521263-B2
Application number	US-201314419924-A
Country	US
Kind code	B2
Filing date	Sep 9, 2013
Priority date	Sep 17, 2012
Publication date	Dec 13, 2016
Grant date	Dec 13, 2016

Abstract

The present document relates to audio communication systems. In particular, the present document relates to the control of the level of audio signals within audio communication systems. A method for leveling a near-end audio signal ( 211 ) using a leveling gain ( 214 ) is described. The near-end audio signal ( 211 ) comprises a sequence of segments, wherein the sequence of segments comprises a current segment and one or more preceding segments. The method comprises determining a nuisance measure ( 416 ) which is indicative of an amount of aberrant voice activity within the sequence of segments of the near-end audio signal ( 211 ); and determining the leveling gain ( 214 ) for the current segment of the near-end audio signal ( 211 ), at least based on the leveling gain ( 214 ) for the one or more preceding segments of the near-end audio signal ( 211 ), and by taking into account—according to a variable degree—an estimate of the level of the current segment of the near-end audio signal ( 211 ); wherein the variable degree is dependent on the nuisance measure ( 416 ).

First claim

The invention claimed is:

1. A method for leveling an audio signal using a leveling gain, the method comprising:
updating the leveling gain for a current segment of the audio signal based on a target level;
wherein the audio signal comprises a sequence of segments that include the current segment;
applying the leveling gain to the current segment of the audio signal; and
repeating the updating and applying for the sequence of segments of the audio signal; wherein the updating of the leveling gain is suspended, subject to determining a pre-determined number of aberrant voice bursts within the audio signal.

2. The method of claim 1 , further comprising:
determining a nuisance measure which is indicative of an amount of aberrant voice activity within the sequence of segments of the audio signal;
increasing the nuisance measure upon detection of an aberrant voice burst within the audio signal; and
suspending the updating of the leveling gain if the nuisance measure exceeds a pre-determined nuisance threshold value.

3. The method of claim 2 , further comprising:
applying a decay factor to the nuisance measure; and
re-enabling the updating of the leveling gain if the nuisance measure falls below the pre-determined nuisance threshold value; wherein the decay factor determines how quickly the updating of the leveling gain is re-enabled.

4. The method of claim 1 , wherein the audio signal represents a near-end audio signal; wherein the sequence of segments comprises the current segment and one or more preceding segments; the method further comprising:
determining a nuisance measure which is indicative of an amount of aberrant voice activity within the sequence of segments of the near-end audio signal; and
determining the leveling gain for the current segment of the near-end audio signal, at least based on the leveling gain for the one or more preceding segments of the near-end audio signal, and by taking into account an estimate of the level of the current segment of the near-end audio signal; wherein the variable degree is dependent on the nuisance measure according to a variable degree.

5. The method of claim 4 , wherein the method further comprises classifying the sequence of segments into voice segments and non-voice segments using voice activity detection.

6. The method of claim 5 , wherein determining the nuisance measure comprises determining a duration of a voice burst of the near-end audio signal, wherein a voice burst comprises one or more successive voice segments; and wherein the nuisance measure depends on the duration of the voice burst.

7. The method of claim 6 , wherein determining the nuisance measure comprises:
if the duration of the voice burst is below a first duration threshold value, increasing the amount of aberrant voice activity indicated by the nuisance measure;
if the duration of the voice burst is above a second duration threshold value, decreasing the amount of aberrant voice activity indicated by the nuisance measure.

8. The method of claim 5 , wherein determining the nuisance measure comprises determining a duration of successive non-voice segments; and wherein the nuisance measure depends on the duration of successive non-voice segments.

9. The method of claim 4 , wherein the nuisance measure is determined at least based on a sequence of segments of a far-end audio signal; wherein the near-end audio signal is derived from an audio signal captured by a microphone of an endpoint of an audio communication system; and wherein the far-end audio signal is to be rendered by a speaker of the endpoint.

10. The method of claim 4 , wherein determining the leveling gain comprises determining whether the nuisance measure exceeds a nuisance threshold; wherein determining the leveling gain comprises, if the nuisance measure exceeds the nuisance threshold, setting the variable degree such that the leveling gain for the current segment of the near-end audio signal is independent of the estimate of the level of the current segment of the near-end audio signal.

11. The method of claim 10 , wherein determining the leveling gain comprises, if the nuisance measure exceeds the nuisance threshold, setting the leveling gain to be equal to the leveling gain for the segment of the near-end audio signal, which is directly preceding the current segment of the near-end audio signal.

12. The method of claim 4 , wherein determining the leveling gain comprises:
determining a probability that the current segment of the near-end audio signal comprises voice using voice activity detection;
weighting the probability with a weighting factor, thereby yielding a weighted probability; wherein the weighting factor depends on the nuisance measure; wherein the weighting factor is such that the weighted probability is decreasing if the amount of aberrant voice activity indicated by the nuisance measure increases; and
if the weighted probability exceeds a probability threshold value, determining the leveling gain for the current segment of the near-end audio signal by taking into account the estimate of the level of the current segment of the near-end audio signal.

13. The method of claim 12 , wherein determining the leveling gain by taking into account the estimate of the level of the current segment of the near-end audio signal comprises:
determining a current estimate of the voice level of the near-end audio signal at least based on a previous estimate of the voice level of the near-end audio signal and based on the estimate of the level of the current segment of the near-end audio signal; and
determining the leveling gain for the current segment of the near-end audio signal such that the current estimate of the voice level of the near-end audio signal corresponds to a target level.

14. The method of claim 12 , wherein determining the leveling gain comprises, if the weighted probability does not exceed the probability threshold value, determine the leveling gain for the current segment of the near-end audio signal without taking into account the estimate of the level of the current segment of the near-end audio signal.

15. The method of claim 4 , wherein determining the leveling gain for the current segment of the near-end audio signal comprises:
updating the leveling gain for the current segment also based on a target level; and
suspending the updating, if the nuisance measure exceeds a nuisance threshold.

16. A computing device for leveling an audio signal using a leveling gain, the computing device comprising one or more computing processors configured to perform:
updating the leveling gain for a current segment of the audio signal based on a target level;
wherein the audio signal comprises a sequence of segments that include the current segment;
applying the leveling gain to the current segment of the audio signal; and
repeating the updating and applying for the sequence of segments of the audio signal; wherein the updating of the leveling gain is suspended, subject to determining a pre-determined number of aberrant voice bursts within the audio signal.

17. The computing device of claim 16 , wherein the one or more computing processors are further configured to perform:
determining a nuisance measure which is indicative of an amount of aberrant voice activity within the sequence of segments of the audio signal;
increasing the nuisance measure upon detection of an aberrant voice burst within the audio signal; and
suspending the updating of the leveling gain if the nuisance measure exceeds a pre-determined nuisance threshold value.

18. The c

Assignees

• Dolby Laboratories Licensing Corp

Inventors

• Shue Yen-Liang
• Gunawan David
• Dickins Glenn N

Classifications

• H04M3/569

  using the instant speaker's algorithm (speech detection per se G10L25/78) · CPC title

• H03G3/32

  the control being dependent upon ambient noise level or sound level · CPC title

• G10L25/78

  Detection of presence or absence of voice signals (switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10) · CPC title

• G10L21/0208 Primary

  Noise filtering · CPC title

• H03G3/301

  the gain being continuously variable · CPC title