adaptive methods for speaker separation in cars
DESCRIPTION
Adaptive Methods for Speaker Separation in Cars. DaimlerChrysler Research and Technology Julien Bourgeois [email protected]. General context. s 2 (t). s 1 (t). +Road Noise spatially diffuse. Several simultaneous speakers (sources) spatially located. x 1 (t). x 4 (t). - PowerPoint PPT PresentationTRANSCRIPT
Adaptive Methods for Speaker Separation in Cars
DaimlerChrysler Research and TechnologyJulien Bourgeois
2
General context
x1(t) x4(t)
Microphones
Goal: provide individual speech input for each passenger
Individual speech flows
s1(t)
s2(t)
+Road Noise spatially diffuse
Several simultaneous speakers (sources) spatially located
Separation
Algorithm
3
General context
x1(t) x4(t)
Microphones Individual speech flows
s1(t)
s2(t)
+Road Noise spatially diffuse
Several simultaneous speakers (sources) spatially located
Separation
Algorithm
Mixing system
Goal: provide individual speech input for each passenger
4
General context
x1(t) x4(t)
Microphones Individual speech flows
s1(t)
s2(t)
+Road Noise spatially diffuse
Several simultaneous speakers (sources) spatially located
Separation
Algorithm
Software
Goal: provide individual speech input for each passenger
5
Plan of the presentation
Overview of existing methods
Supervised/Informed separation vs. Blind separation
Blind separation and prior spatial information
Conclusion and future work
6
Existing methods: CASA vs. Multichannel Techniques
CASA: 1 microphone separation
Heuristics based on an analysis of human auditory system
Requires a lot of data (training of parameters)
Multi-microphones techniques: Speech moves much faster than…
the coherence relating two (or more) microphones.
7
Existing Methods: Beamforming
Beamforming: Prior information on target position
Constrain the response in the direction of interest
Minimize the output power
Problem of target cancellation if prior
spatial info is not perfect.Filters
Direction of interest
output
8
Existing methods: Blind Source Separation
Blind Source Separation (BSS) First applications to speech separation at the end of the
90’s
Only requirement: statistically independent sources
Difficult optimization problem: maximizing a nonlinear function (independence measure).
With many microphones, target cancellation can also appear.
Permutation ambiguity.
Acoustic
Mixing
BSSSources
DependentObservations
IndependentOutputs
9
The question is…
Is it possible to merge Beamforming and BSS, and combine their advantages?
In cars, prior knowlegde on speaker positions, separate blindly is suboptimal.
10
Blind separation and prior spatial information
Initialisation of BSS according to speakers positions helps optimisation procedure a lot.
Solve permutations problem solved
Target cancellation problem solved
Prior info : positions
Acoustic
Mixing
BSSSources
DependentObservations
IndependentOutputs
11
BSS is not that blind…
BSS performances depends dramatically on the type of mixing Strictly causal
12
BSS is not that blind…
BSS performances depends dramatically on the type of mixing Strictly causal
Non strictly causal
13
Beamforming is not that informed…
Perfect prior spatial information is actually not necessary:
Target cancellation problem can be solved if one can detect activity/silences of each speaker.
The detection problem is strongly related with IDIAP smart meeting room projects.
Filters
Direction of interest
output
14
Conclusion and future works
Combining BSS with a beamformer is not gainful.
We may inform BSS efficiently in the case of non-causal mixings
(algorithmic rotation of the microphone array)
15
Conclusion and future works
Combining BSS with a beamformer is not gainful.
We may inform BSS efficiently in the case of non-causal mixings
(algorithmic rotation of the microphone array)
16
Thank you!