Method and system of audio retrieval and source separation
A method and a system of audio retrieval and source separation are described. The method comprises the steps of: receiving a textual query; retrieving a preliminary audio sample from an auxiliary audio database; retrieving a target audio sample from a target audio database; and separating the retrieved target audio sample into a plurality of audio source signals. The corresponding system comprises an input unit, a storing unit and a processing unit to implement the method.
1. A method comprising:
receiving a textual query;
retrieving a preliminary audio sample from an auxiliary audio database by matching the textual query with semantic information associated to the auxiliary audio database;
performing one of
retrieving a target audio sample from a target audio database by matching the preliminary audio sample with the target audio database; and performing an audio-source separation technique on the retrieved target audio sample for separating the retrieved target audio sample into a plurality of audio source signals, wherein retrieving the target audio sample from the target audio database and performing the audio-source separation technique on the retrieved target audio sample are performed jointly, retrieving the target audio sample from the target audio database including comparing the target audio sample with a negative data set; and
separating a target source signal of an audio mixture, said target source signal being described by said textual query, by matching the preliminary audio sample with the audio mixture.
2. The method of claim 1 , comprising reviewing the preliminary audio sample retrieved from the auxiliary audio database.
3. The method of claim 2 , wherein the reviewing includes segmenting or filtering the retrieved preliminary audio sample, and is performed upon receipt of a user input.
4. The method of claim 1 , wherein the negative data set is generated from the auxiliary database or a second auxiliary database.
5. The method of claim 1 , wherein the audio source signals of the retrieved target audio sample include at least one signal corresponding to the textual query and at least one signal not corresponding to the textual query.
6. The method of claim 1 , comprising iterating some or all of the method.
7. A system comprising:
an input unit for receiving a textual query;
a processing unit for retrieving a preliminary audio sample from an auxiliary audio database by matching the textual query with semantic information associated to the auxiliary audio database, and one of:
retrieving a target audio sample from a target audio database by matching the preliminary audio sample with the target audio database and performing an audio-source separation technique on the retrieved target audio sample for separating the retrieved target audio sample into a plurality of audio source signals, wherein retrieving the target audio sample from the target audio database and performing the audio-source separation technique on the retrieved target audio sample are performed jointly, retrieving the target audio sample from the target audio database by the processing unit including comparing the target audio sample with a negative data set stored in a storing unit of said system; and
performing an audio-source separation technique on an audio source mixture for separating a target source signal described by said textual query from said audio source mixture by matching the preliminary audio sample with the audio mixture.
8. The system of claim 7 , wherein the negative data set is generated from the auxiliary audio database or a second auxiliary database.
9. The system of claim 7 , wherein the processing unit reviews the preliminary audio sample retrieved from the auxiliary audio database upon receipt of a user input in the input unit.
10. A non-transitory computer readable medium having stored therein instructions for performing the method according to claim 1 .
11. The method of claim 1 , wherein the auxiliary audio database includes an internet search engine.
12. The system of claim 7 , wherein the auxiliary audio database includes an internet search engine.
13. The system of claim 7 , wherein the system includes a storing unit for storing the target audio database.
14. The method of claim 1 , where said target audio database is a database without semantic information associated thereto.
15. The method of claim 1 , where said retrieving of said target audio sample from said target audio database is performed by an examplar-base matching between of said preliminary audio sample with the target audio database.
16. The method of claim 1 , where said matching of the preliminary audio sample with the audio source mixture is an examplar-base matching.