Method and apparatus for expanding audio data
Systems implementing the invention allow a user to time stretch an audio track without changing the pitch of the sound, and to produce optimal audible qualities of the output signal. The approach utilized in the invention relies on providing several time stretching methods, each one of which is selected based on one or more criteria of the audio data properties. One method relies on crossfading pairs of segments of audio data while running one segment backward every other repetition. The second time stretching method detects inaudible segments and inserts longer periods of audible data within those segments. The third method utilizes a reverb to create a reverb segment that is played after the original segment.
1. A method for time stretching audio data without changing the pitch comprising:
obtaining at least one audio data stream;
obtaining at least one energy property representation of said at least one audio data stream;
obtaining at least one optimal input segment for time stretching using said at least one energy property representation;
defining a first segment and a second segment that at least overlap said optimal input segment; and
generating an output segment by sequentially crossfading said first segment and said second segment;
wherein sequentially crossfading comprises a first crossfading of said first segment and said second segment while reversing the sense of said first segment and a second crossfading of said first segment and said second segment while reversing the sense of said second segment.
2. The method of claim 1 , wherein said obtaining at least one energy property representation further comprises computing a square of the amplitude of data samples in said audio stream.
3. The method of claim 1 , wherein said obtaining said at least one optimal input segment further comprises obtaining a plurality of adjacent segments in said audio stream.
4. The method of claim 1 , wherein said defining said first segment and said second segment further comprises defining a plurality of boundaries associated with said first segment and said second segment.
5. The method of claim 1 , wherein said defining said plurality of boundaries further comprises defining boundaries for copying unedited audio segments.
6. The method of claim 1 , wherein said crossfading said first segment and said second segment further comprises computing a fade-out coefficient and a fade-in coefficient.
7. The method of claim 6 , wherein said crossfading said first segment and said second segment further comprises computing a first product of said first segment with said fade-out coefficient and a second product of said second segment and said fade-in coefficient.
8. The method of claim 7 , wherein said crossfading said first segment and said second segment further comprises summing said first product and said second product.
9. The method of claim 1 , wherein said reversing the sense of said at least one of said first segment and said second segment further comprises running an index from the end of said at least one of said first segment and said second segment.
10. The method of claim 1 , wherein said sequentially crossfading further comprises copying at least a portion of unedited data from said data stream to said output segment.
11. A computer-readable medium carrying one or more sequences of instructions executable on a computer for time stretching audio data without changing the pitch, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:
obtaining at least one audio data stream;
obtaining at least one energy property representation of said at least one audio data stream;
obtaining at least one optimal input segment for time stretching using said at least one energy property representation;
defining a first segment and a second segment that at least overlap said optimal input segment; and
generating an output segment by sequentially crossfading said first segment and said second segment;
wherein sequentially crossfading comprises a first crossfading of said first segment and said second segment while reversing the sense of said first segment and a second crossfading of said first segment and said second segment while reversing the sense of said second segment.
12. An apparatus comprising:
a network interface;
a memory; and
one or more processors connected to the network interface and the memory, the one or more processors configured for
obtaining at least one audio data stream;
obtaining at least one energy property representation of said at least one audio data stream;
obtaining at least one optimal input segment for time stretching using said at least one energy property representation;
defining a first segment and a second segment that at least overlap said optimal input segment; and
generating an output segment by sequentially crossfading said first segment and said second segment;
wherein sequentially crossfading comprises a first crossfading of said first segment and said second segment while reversing the sense of said first segment and a second crossfading of said first segment and said second segment while reversing the sense of said second segment.