Audio encoding apparatus and methods

• US 6,266,644 B1
• Filed: 09/26/1998
• Issued: 07/24/2001
• Est. Priority Date: 09/26/1998
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Liquid Audio Inc
• Inventors: Levine, Scott Nathan
• Primary Examiner(s): Dorvil, Richemond
• Application Number:
  US09/161,198
• Time in Patent Office: 1,032 Days
• Field of Search: 704/500, 704/501, 704/502, 704/503, 704/504, 704/226, 704/227, 704/228, 704/203, 704/209, 704/200.1
• US Class Current: 704/503
• CPC Class Codes: G10L 19/0212 using orthogonal transformation
• CPC Class Codes: G10L 19/093 using sinusoidal excitation models
• CPC Class Codes: G10L 19/20 using sound class specific coding, hybrid encoders or object based coding

Abstract

An audio encoding system includes quantization, phase matching, compressed domain processing and other improvements enabling high fidelity, low bit rate encoding systems to be formed. In a preferred embodiment a novel transient detector is used to divide an audio source into transient, low frequency non-transient and high frequency non-transient regions. Low frequency non-transients are sinusoid modeled and the residual is low frequency noise modeled. Transients are transform coded and high frequency non-transients and transform coded residual is high frequency noise modeled. The preferred embodiment also includes novel sinusoid, transform coded and noise quantization, among other methods for providing high fidelity data reduction and for interfacing sinusoid modeling and transform coding. Compressed domain time compression and expansion are further achieved with no detrimental effect on transients.

Claims

1. An audio encoder comprising:

at least one parametric modeler;

at least one non-parametric modeler coupled to said parametric modeler; and

at least one encoder which is coupled to the at least one parametric modeler and to the at least one non-parametric modeler and which produces encoded audio data from an audio source.

2. An audio encoder according to claim 1, wherein said at least one non-parametric modeler comprises a transform coder.

3. An audio encoder according to claim 1, wherein said at least one parametric modeler is selected from a group comprising sinusoidal modelers and noise modelers.

4. An audio encoder according to claim 3, wherein said at least one parametric modeler comprises a sinusoidal modeler, a low frequency noise modeler and a high frequency noise modeler.

5. An audio encoder according to claim 1, wherein said at least one parametric modeler comprises a sinusoidal modeler, a low frequency noise modeler and a high frequency noise modeler, and said non-parametric modeler comprises a transform coder.

6. An audio encoder according to claim 1, wherein operational parameters of said audio encoder are predetermined.

7. An audio encoder according to claim 1, wherein operational parameters of said audio encoder are modifiable.

8. An audio encoder according to claim 7, wherein operational parameters of said audio encoder are modifiable in accordance with automatically sensed characteristics of said audio source.

9. An audio encoder according to claim 7, wherein operational parameters of said audio encoder are modifiable in accordance with user input to said audio encoder.

10. An audio encoding method comprising:

(a) receiving an audio source;

(b) modeling non-transient portions of said audio source using at least one parametric modeler;

(c) modeling transient portions of said audio source using at least one nonparametric modeler; and

(d) producing encoded audio data from the audio source in accordance with the modeling of (b) and (c).

11. A method according to claim 10, wherein said modeling step (b) comprises:

sinusoidally modeling a low frequency band of said non-transient portions of said audio source, thereby producing sinusoidally modeled audio;

quantizing said sinusoidally modeled audio, thereby producing a sinusoidal residual;

low frequency noise modeling said sinusoidal residual; and

high frequency noise modeling a high frequency band of said non-transient portions of said audio source.

12. A method according to claim 10, wherein said modeling step (c) comprises:

substantially transform coding said transient portions of said audio source, thereby producing transform coded audio;

quantizing said transform coded audio, thereby producting a transient residual; and

high frequency noise modeling said transient residual.

13. Audio data encoded according to the method of claim 10.