Google Inc.

Andrew Allen

Jan K Skoglund

Willem Bastiaan Kleijn

Sze Chie Lim

Marcin Gorzel

Jamieson Brettle

Andrew Hines

Miroslaw Narbutt

Techniques of rendering sound for a listener involve determining, at a first position, a set of directions from which some specified fraction of the sound field energy originates and reconstruct the sound field at a second position based on the set of directions. Along these lines, a sound rendering computer may receive sound data representing sound emanating from positions on a sphere centered on a microphone. Nevertheless, only a small fraction of these points on the sphere contribute much of the sound field energy directed at the listener. The sound rendering computer determines such directions using a directional decomposition in which those directions having the highest fraction of the sound energy are identified. When the listener moves away from the microphone's position, the sound rendering computer, after identifying such directions, applies an amplitude and phase factor to the sound fields corresponding to those directions.

Ambisonics sound field navigation using directional decomposition and path distance estimation

A method of encoding sound objects includes receiving a set of monophonic sound inputs. Each of the set of monophonic sound inputs includes position and orientation information of a sound object relative to a source position. The set of monophonic sound inputs are encoded into a higher order ambisonic (HOA) sound field in a spherical harmonics domain based on a spherical harmonics dataset including a subset of all spherical harmonic coefficients for a given subset of azimuth and elevation angles. Some embodiments include decoding the HOA sound field to generate a set of loudspeaker signals.

Fast and memory efficient encoding of sound objects using spherical harmonic symmetries

A computing device includes a processor and a memory. The processor is configured to generate spectrograms, for example, using short-time Fourier transform, for a plurality of channels of reference and test ambisonic signals. In some implementations, the test ambisonic signal may be generated by decoding an encoded version of the reference ambisonic signal. The processor is further configured to compare, for each of the plurality of channels of a reference ambisonic signal, at least a patch associated with a channel of the reference ambisonic signal with at least a corresponding patch of a corresponding channel of the test ambisonic signal and determine a localization accuracy of the test ambisonic signal based on the comparison. In some implementations, the comparing may be based on phaseograms of the reference and test ambisonic signals.

Objective quality metrics for ambisonic spatial audio

Techniques of performing binaural rendering involve combining left and right HRTFs from speakers symmetrically placed with respect to a forward axis of a listener's head. Because of the symmetry of the listening geometry, a decoded loudspeaker signal resulting from each speaker may be decomposed into symmetric and antisymmetric components according to whether a spherical harmonic associated with a respective ambisonic channel is symmetric or antisymmetric about the forward axis. The result involves rendering the left and right headphone speakers using the same sum and difference HRTFs rather than a different pair of HRFTs for each headphone speaker.

Symmetric binaural rendering for high-order ambisonics

Techniques of performing involve providing interactive audio in addition to ambisonic audio in stereo tracks selected according to the occurrence of events in a media delivery system. For example, a user of a VR system observes a virtual environment that contains many virtual objects. The user may experience binaurally rendered audio played over N ambisonic channels from any number of virtual loudspeakers. In addition, the user may also activate another audio source by positioning his/her head at a certain angle, e.g., to look at a particular virtual object. As a specific example, when the user looks at a picture of a person, an audio track may play over a pair of stereo channels N+1 and N+2. Because they are stereo channels, there is no need to perform convolutions with HRTFs. In this way, audio may be provided for all virtual objects in the virtual environment with a small computational overhead.

Ambisonic audio with non-head tracked stereo based on head position and time

Techniques of rendering sound for a listener involve producing, as the amplitude of each of the source driving signals, a sum of two terms: a first term based on a solution sto the equation b=A&#xb7;s, and a second term based on a projection of a specified vector &#x15d; onto the nullspace of A, &#x15d; not being a solution to the equation b=A&#xb7;s. Along these lines, in one example, the first term is equivalent to a Moore-Penrose pseudoinverse, e.g., A(AA)&#xb7;b. In general, any solution to the equation b=A&#xb7;s is satisfactory. The specified vector that is projected onto the nullspace of A is defined to reduce the coherence of the net sound field. Advantageously, the resulting operator is both linear time-invariant and idempotent so that the sound field may be faithfully reproduce both inside the RSF and at a sufficient range outside the RSF to cover a human head.

Incoherent idempotent ambisonics rendering

Techniques of performing binaural rendering involve separating symmetric and antisymmetric terms in the total output rendered in the ears of a listener. Along these lines, a sound field includes a set of sound field weights corresponding to spherical harmonic (SH) functions in a SH expansion of the sound field. In addition, an aggregate head-related transfer function (HRTF) includes a set of HRTF weights that correspond to a SH function. An HRTF weight may be generated from aggregating products of an HRTF at each of a set of loudspeaker positions and a SH function to which the HRTF weight corresponds at that loudspeaker position. The rendered sound field in one of the ears of the listener would be, when the sound field and HRTF is a function of frequency, a sum of the products of corresponding sound field weights and HRTF weights. One may save much computation by grouping the products into symmetric terms and antisymmetric terms. The rendered sound field in, say, the left ear is the sum over each loudspeaker position of the sum of the symmetric terms and antisymmetric terms for that loudspeaker position. Accordingly, because the head of the listener is assumed symmetric about the forward axis, the rendered sound field in the right ear is the sum over each loudspeaker position of the difference between the symmetric terms and antisymmetric terms for that loudspeaker position.

Symmetric spherical harmonic HRTF rendering

Techniques of performing binaural rendering involve generating separate locations of virtual sources on the sphere for each ear of a listener. Along these lines, consider a set of actual audio sources that are not equidistant from a central point. To provide a listener with ambisonic audio, a sphere is defined with the listener at its center. When a source is not on the surface of the sphere, respective rays from the source to each of the listener's ears may not intersect the sphere at the same point. Rather, to provide a more accurate representation of the actual source, virtual loudspeakers are placed at each of the sphere intersections, a first virtual loudspeaker propagating audio to the left ear, a second virtual loudspeaker propagating audio to the right ear.

Growing community of inventors

San Jose, CA, United States of America

Andrew Allen