Content-aware audio level management
AssigneeDolby Laboratories Licensing Corporation
InventorsShaofan Yang, Kai Li, Qianqian Fang
Life of this patent
Estimating the term…
Abstract
Described herein is a method of performing content-aware audio processing for an audio signal comprising a plurality of audio components of different types. The method includes source separating the audio signal into at least a voice-related audio component and a residual audio component. The method further includes determining a dynamic audio gain based on the voice-related audio component and the residual audio component. The method also includes performing audio level adjustment for the audio signal based on the determined audio gain. Further described are corresponding apparatus, programs, and computer-readable storage media.
Claims33
A method of performing content-aware audio processing for an audio signal comprising a plurality of audio components of different types, the method comprising:
- source separating the audio signal into at least a voice-related audio component and a residual audio component;
- determining a dynamic audio gain based on the voice-related audio component and the residual audio component, wherein the determination of the dynamic audio gain involves identifying whether the audio signal relates to professionally generated content or user generated content;
- and
- performing audio level adjustment for the audio signal based on the determined dynamic audio gain.
A method of performing content-aware audio processing for an audio signal comprising a plurality of audio components of different types, the method comprising:
- source separating the audio signal into at least a voice-related audio component and a residual audio component;
- determining a dynamic audio gain based on the voice-related audio component and the residual audio component, wherein the determination of the dynamic audio gain involves identifying whether the residual audio component relates to noise or not;
- and
- performing audio level adjustment for the audio signal based on the determined dynamic audio gain,
- wherein the residual audio component is represented in a plurality of time-domain blocks, and
- wherein identifying whether the residual audio component relates to noise or not comprises:
- determining a first confidence score indicative of whether a current block of the residual audio component relates to noise or not based on the current block and at least one time-adjacent block of the residual audio component.
A method of performing content-aware audio processing for an audio signal comprising a plurality of audio components of different types, the method comprising:
- source separating the audio signal into at least a voice-related audio component and a residual audio component;
- determining a dynamic audio gain based on the voice-related audio component and the residual audio component;
- and
- performing audio level adjustment for the audio signal based on the determined dynamic audio gain, wherein the audio level adjustment is performed by applying the determined audio gain in the time-frequency domain according to
- Z[k,t]≡G[k,t]{X[k,t]+(1−Py[t]Pz[t])Y[k,t]},
- wherein Z[k, t] denotes the level-adjusted audio signal, X[k, t] denotes the voice-related audio component in the audio signal, Y[k, t] denotes the residual audio component in the audio signal, PY[t] denotes a confidence score indicative of whether the residual audio component is identified to relate to noise, PZ[t] denotes a confidence score indicative of whether the audio signal is identified to relate to user generated content, k denotes the frequency index, and t denotes the time index.
Dependencies are inferred from the claim text.