Decoding-Time Prediction of Non-Verbalized Tokens
AssigneeMultimodal Technologies, LLC
InventorsFritsch Juergen, Anoop Deoras, Detlef Koll
Life of this patent
Estimating the term…
Abstract
Non-verbalized tokens, such as punctuation, are automatically predicted and inserted into a transcription of speech in which the tokens were not explicitly verbalized. Token prediction may be integrated with speech decoding, rather than performed as a post-process to speech decoding.
Claims25
A computer-implemented method comprising:
- (A) decoding a first portion of an audio signal into a first word in a token stream;
- (B) after (A), inserting a punctuation mark into the token stream at a position after the first word;
- and
- (C) after (B), decoding a second portion of the audio signal into a second word in the token stream at a position after the punctuation mark.
A computer program product tangibly storing computer program instructions executable by a computer processor, the computer program instructions comprising instructions for:
- (A) decoding a first portion of an audio signal into a first word in a token stream;
- (B) after (A), inserting a punctuation mark into the token stream at a position after the first word;
- and
- (C) after (B), decoding a second portion of the audio signal into a second word in the token stream at a position after the punctuation mark.
A computer-implemented method comprising:
- (A) using a language model to decode a first portion of an audio signal into a first word in a token stream;
- (B) after (A), selecting a token based on the language model and the first word;
- (C) inserting the selected token into the token stream at a position after the first word;
- and
- (D) after (B), using the language model to decode a second portion of the audio signal into a second word in the token stream at a position after the punctuation mark.
A computer program product tangibly storing computer program instructions executable by a computer processor, the computer program instructions comprising instructions for:
- (A) using a language model to decode a first portion of an audio signal into a first word in a token stream;
- (B) after (A), selecting a token based on the language model and the first word;
- (C) inserting the selected token into the token stream at a position after the first word;
- and
- (D) after (B), using the language model to decode a second portion of the audio signal into a second word in the token stream at a position after the punctuation mark.
Dependencies are inferred from the claim text.