System and method for phonetic search over speech recordings

1. A computerized-method of searching for an element in speech related documents, the method comprising:

transcribing, by a controller, a set of digital speech recordings stored in a speech recording database to a set of digital phoneme strings and including the phoneme strings in a set of digital phonetic transcriptions;

reverse-indexing, by the controller, using one or more parser rules, the phonetic transcriptions according to one or more phonemes and storing the reverse-indexed transcriptions in an inverted index database including a mapping to documents such that the one or more phonemes can be used as a search key for searching in the phonetic transcriptions for a phoneme string;

transcribing, by the controller, a textual search term into a set of search phoneme strings; and

using, by the controller, the set of search phoneme strings to search for the textual search term in the set of phonetic transcriptions and produce a list of the documents ranked according to the match of the phoneme strings to the documents.

2. The method of claim 1 , comprising, prior to reverse-indexing the phonetic transcriptions, performing at least one of:

removing a phoneme string from the set of phonetic transcriptions, and

replacing a phoneme string in the set of phonetic transcriptions.

3. The method of claim 1 , comprising:

producing N-Best phonetic transcriptions based on an input speech recording; and

inverse indexing the N-Best phonetic transcriptions to produce a set of N phonetic transcriptions searchable using the textual search term.

4. The method of claim 1 , comprising:

dividing the set of phonetic transcriptions into a plurality of short sub-documents; and

indexing the sub-documents according to one or more phonemes such that a phoneme can be used as a search key for searching in the sub-documents for the phoneme.

5. The method of claim 1 , comprising constructing a searchable term based on an index proximity of two or more phonemes identified in the set of phonetic transcriptions.

6. The method of claim 1 , comprising:

transcribing the textual search term into a set of search phoneme strings based on a respective set of pronunciations, and

simultaneously searching for the set of search phoneme strings in the set of phonetic transcriptions.

7. The method of claim 1 , comprising including, in the set of search phoneme strings, at least one phoneme string based on a pre-configured distance from the textual search term.

8. The method of claim 1 , comprising:

statistically calculating a probability of a recognition error for a phoneme; and

based on relating a fuzziness parameter value to the probability, selecting to include or exclude a phonetic transcription in a result of searching for an element in the set of phonetic transcriptions.

9. The method of claim 1 , comprising:

statistically calculating a probability of a recognition error for a phoneme; and

based on relating a fuzziness parameter value to the probability, selecting to include the phoneme in the set of search phoneme strings.

10. The method of claim 1 , comprising, transcribing the textual search term into a set of search phoneme strings and using the set of search phoneme strings to conduct a set of searches.

11. The method of claim 1 , comprising, identifying, in the set of phonetic transcriptions, phonetic transcriptions that include an element that is similar to the searched element.

12. The method of claim 1 , comprising:

using the set of search phoneme strings to search for a set of elements in the set of phonetic transcriptions; and

identifying, in the set of phonetic transcriptions, phonetic transcriptions that include a sub-set of elements included in a set of elements.

13. A system of searching for an element in speech related documents, the system comprising:

a non-transitory memory; and

a hardware controller configured to:

transcribe a set of speech recordings stored in a speech recording database to a set of phoneme strings and include the phoneme strings in a set of phonetic transcriptions,

reverse-index, using one or more parser rules, the phonetic transcriptions according to one or more phonemes, and store the reverse-indexed transcriptions in an inverted index database including a mapping to documents such that the one or more phonemes can be used as a search key for searching in the phonetic transcriptions for a phoneme string,

transcribe a textual search term into a set of search phoneme strings, and

use the set of search phoneme strings to search for an element in the set of phonetic transcriptions and produce a list of the documents ranked according to the match of the phoneme strings to the documents.

14. The system of claim 13 , wherein the controller is configured to, prior to reverse-indexing the phonetic transcriptions, performing at least one of:

removing a phoneme string from the set of phonetic transcriptions, and

replacing a phoneme string in the set of phonetic transcriptions.

15. The system of claim 13 , wherein the controller is configured to:

produce N-Best phonetic transcriptions based on an input speech recording; and

inverse index the N-Best phonetic transcriptions to produce a set of N phonetic transcriptions searchable using the textual search term.

16. The system of claim 13 , wherein the controller is configured to:

divide the set of phonetic transcriptions into a plurality of short sub-documents; and

index the sub-documents according to one or more phonemes such that a phoneme can be used as a search key for searching in the sub-documents for the phoneme.

17. The system of claim 13 , wherein the controller is configured to construct a searchable term based on an index proximity of two or more phonemes identified in the set of phonetic transcriptions.

18. The system of claim 13 , wherein the controller is configured to:

transcribe the textual search term into a set of search phoneme strings based on a respective set of pronunciations, and

simultaneously search for the set of search phoneme strings in the set of phonetic transcriptions.

19. The system of claim 13 , wherein the controller is configured to include, in the set of search phoneme strings, at least one phoneme string based on a pre-configured distance from the textual search term.

20. The system of claim 13 , wherein the controller is configured to:

statistically calculate a probability of a recognition error for a phoneme; and

based on relating a fuzziness parameter value to the probability, select to include or exclude a phonetic transcription in or from a result of searching for an element in the set of phonetic transcriptions.