Method for projected regularization of audio data
AssigneeSteinberg Media Technologies GmbH
InventorsJean-Baptiste Jacques Guillaume Rolland, Yvan Grabit
Life of this patent
Estimating the term…
Abstract
A method of modification of audio data to improve the quality of the audio modification or reconstruction or improves the speed of such reconstruction or modification and produces more realistic audio data. Realistic audio data is audio data that is generated in natural events like talking or singing or a vehicle passing by and is not generated only by artificially constructing audio data like in a synthesizer. This will lead to audio data that will be perceived more likely as natural or unmodified audio signal when being played back to human beings. The method involves modification of some part of transformed audio data, especially phase data.
Claims19
A method of processing first audio data comprising:
- a. transforming the first audio data using a first transformation generating time-frequency domain audio representation data, especially spectrogram data, having a first magnitude and first phase data and using the first magnitude and first phase data as a first iterated magnitude and first iterated phase data;
- b. iteratively changing the transformed first audio data by looping the following steps:
- i. modifying the first iterated magnitude and first iterated phase data to obtain modified magnitude and modified phase data;
- ii. at least in every 5th loop pass performing the following steps:
- 1. retransforming the modified magnitude and modified phase data to generate complex data;
- 2. calculating more real data from the complex data;
- 3. transforming the more real data to generate second phase data and using a combination of the modified magnitude and second phase data with the first magnitude and first phase data as new iterated magnitude and new iterated phase data or
- using a combination of the more real data with a transformation of the first magnitude and first phase data as new data and transforming the new data generating new iterated magnitude and new iterated phase data
- and in every other loop pass
- 3a. using a combination of the modified magnitude and modified phase data with the first magnitude and first phase data as new iterated magnitude and iterated phase data
- c. retransforming the iterated magnitude and iterated phase data to generate modified audio data.
A method of processing first audio data, comprising:
- providing an audio processing device including at least one processor and at least one memory;
- wherein the at least one processor is programmed to perform an algorithm that processes first audio data and improves the quality of audio data modification or gives the audio data a more realistic or natural sound;
- the algorithm including:
- transforming the audio data in the audio processing device through a series of loop passes;
- while
- at least in every 5th loop pass the audio data is changed to become more real.
Dependencies are inferred from the claim text.