Commits · Xenobd/whisper.cpp

Try to improve the token sampling strategy (#193)

8311a60
unverified

ggerganov commited on Dec 2, 2022

whisper : add mechanism for aborting the whisper_full() computation

d311de4

ggerganov commited on Nov 27, 2022

whisper.objc : add real-time processing (#97)

11bb554
unverified

ggerganov commited on Nov 26, 2022

whisper.objc : fix build warnings

8d1f7e9
unverified

ggerganov commited on Nov 26, 2022

whisper.wasm : do not block page while processing (close #86)

d0b1d9e
unverified

ggerganov commited on Nov 25, 2022

refactoring : more readable code

5ef0168
unverified

ggerganov commited on Nov 25, 2022

stream : "-kc" now enables context keeping from previous segment (#90)

28726dd
unverified

ggerganov commited on Nov 22, 2022

Prompt previous tokens for streaming (#163)

8ad3dbf
unverified

meakbiyik commited on Nov 22, 2022

stream : add "audio_ctx" parameter

6adc1fe

ggerganov commited on Nov 20, 2022

stream : add "max_tokens" parameter

e48ba5c

ggerganov commited on Nov 20, 2022

stream : add "single_segment" option

a265bfa

ggerganov commited on Nov 20, 2022

stream : partial encoder experiments

a2015c0

ggerganov commited on Nov 11, 2022

whisper : add option to speed up the audio tempo by x2

bec875e

ggerganov commited on Nov 12, 2022

ref #22 : add "duration" option

097e96a
unverified

ggerganov commited on Nov 7, 2022

whisper : token-level timestamp refactoring (#49, #120)

e4f586b
unverified

ggerganov commited on Nov 2, 2022

main : add option for word-leve timestamps (very experimental)

b777dde

ggerganov commited on Oct 30, 2022

close #113 : fix struct whisper_token_data

ba9fc0a
unverified

ggerganov commited on Oct 30, 2022

parallel : print time of audio boundaries + fix timings

84ba527

ggerganov commited on Oct 29, 2022

parallel : working

d5d2d41

ggerganov commited on Oct 29, 2022

main : fix sampling time + add max_context parameter

9ed1355

ggerganov commited on Oct 29, 2022

parallel : adding tool for parallel transformer inference

cafe46d

ggerganov commited on Oct 29, 2022

ggml : add system info functions

7a4a7a1

ggerganov commited on Oct 25, 2022

whisper : add new-segment callback

83926f7

ggerganov commited on Oct 22, 2022

wip : experimental color coding of tokens based on probabilities

6f2f468

ggerganov commited on Oct 21, 2022

ref #52 : improve greedy sampling strategy

07043b9
unverified

ggerganov commited on Oct 18, 2022

close #32 : add comment about thread-safety of the C-style API

a3ee28a
unverified

ggerganov commited on Oct 18, 2022

ref #57, #62, #63 : remove unions in C-api + remove designated initializers

99f2636
unverified

ggerganov commited on Oct 18, 2022

ref #35 : add <stdbool.h> to whisper.h

0ab5514
unverified

ggerganov commited on Oct 10, 2022

ref #9 : add API documentation in whisper.h

ce2cd6d

ggerganov commited on Oct 8, 2022

Adding sanitizer tests

6113fbd

ggerganov commited on Oct 8, 2022

ref #10 : option to keep context in "stream" example

606a6bc
unverified

ggerganov commited on Oct 7, 2022

ref #16, #22 : add "offset" argument

35a7fcb
unverified

ggerganov commited on Oct 7, 2022

Improve result printing

9a513f5

ggerganov commited on Oct 4, 2022

Extend C-style API with full inference methods

dabc473

ggerganov commited on Oct 4, 2022

Initial C-style interface for whisper.cpp

c0943fb

ggerganov commited on Oct 4, 2022

Spaces:

Duplicated from natasa365/whisper.cpp

Xenobd
/

whisper.cpp

Running

Commit History

Try to improve the token sampling strategy (#193)

8311a60
unverified

whisper : add mechanism for aborting the whisper_full() computation

d311de4

whisper.objc : add real-time processing (#97)

11bb554
unverified

whisper.objc : fix build warnings

8d1f7e9
unverified

whisper.wasm : do not block page while processing (close #86)

d0b1d9e
unverified

refactoring : more readable code

5ef0168
unverified

stream : "-kc" now enables context keeping from previous segment (#90)

28726dd
unverified

Prompt previous tokens for streaming (#163)

8ad3dbf
unverified

stream : add "audio_ctx" parameter

6adc1fe

stream : add "max_tokens" parameter

e48ba5c

stream : add "single_segment" option

a265bfa

stream : partial encoder experiments

a2015c0

whisper : add option to speed up the audio tempo by x2

bec875e

ref #22 : add "duration" option

097e96a
unverified

whisper : token-level timestamp refactoring (#49, #120)

e4f586b
unverified

main : add option for word-leve timestamps (very experimental)

b777dde

close #113 : fix struct whisper_token_data

ba9fc0a
unverified

parallel : print time of audio boundaries + fix timings

84ba527

parallel : working

d5d2d41

main : fix sampling time + add max_context parameter

9ed1355

parallel : adding tool for parallel transformer inference

cafe46d

ggml : add system info functions

7a4a7a1

whisper : add new-segment callback

83926f7

wip : experimental color coding of tokens based on probabilities

6f2f468

ref #52 : improve greedy sampling strategy

07043b9
unverified

close #32 : add comment about thread-safety of the C-style API

a3ee28a
unverified

ref #57, #62, #63 : remove unions in C-api + remove designated initializers

99f2636
unverified

ref #35 : add <stdbool.h> to whisper.h

0ab5514
unverified

ref #9 : add API documentation in whisper.h

ce2cd6d

Adding sanitizer tests

6113fbd

ref #10 : option to keep context in "stream" example

606a6bc
unverified

ref #16, #22 : add "offset" argument

35a7fcb
unverified

Improve result printing

9a513f5

Extend C-style API with full inference methods

dabc473

Initial C-style interface for whisper.cpp

c0943fb

Commit History

Try to improve the token sampling strategy (#193) 8311a60 unverified

whisper : add mechanism for aborting the whisper_full() computation d311de4

whisper.objc : add real-time processing (#97) 11bb554 unverified

whisper.objc : fix build warnings 8d1f7e9 unverified

whisper.wasm : do not block page while processing (close #86) d0b1d9e unverified

refactoring : more readable code 5ef0168 unverified

stream : "-kc" now enables context keeping from previous segment (#90) 28726dd unverified

Prompt previous tokens for streaming (#163) 8ad3dbf unverified

stream : add "audio_ctx" parameter 6adc1fe

stream : add "max_tokens" parameter e48ba5c

stream : add "single_segment" option a265bfa

stream : partial encoder experiments a2015c0

whisper : add option to speed up the audio tempo by x2 bec875e

ref #22 : add "duration" option 097e96a unverified

whisper : token-level timestamp refactoring (#49, #120) e4f586b unverified

main : add option for word-leve timestamps (very experimental) b777dde

close #113 : fix struct whisper_token_data ba9fc0a unverified

parallel : print time of audio boundaries + fix timings 84ba527

parallel : working d5d2d41

main : fix sampling time + add max_context parameter 9ed1355

parallel : adding tool for parallel transformer inference cafe46d

ggml : add system info functions 7a4a7a1

whisper : add new-segment callback 83926f7

wip : experimental color coding of tokens based on probabilities 6f2f468

ref #52 : improve greedy sampling strategy 07043b9 unverified

close #32 : add comment about thread-safety of the C-style API a3ee28a unverified

ref #57, #62, #63 : remove unions in C-api + remove designated initializers 99f2636 unverified

ref #35 : add <stdbool.h> to whisper.h 0ab5514 unverified

ref #9 : add API documentation in whisper.h ce2cd6d

Adding sanitizer tests 6113fbd

ref #10 : option to keep context in "stream" example 606a6bc unverified

ref #16, #22 : add "offset" argument 35a7fcb unverified

Improve result printing 9a513f5

Extend C-style API with full inference methods dabc473

Initial C-style interface for whisper.cpp c0943fb

Try to improve the token sampling strategy (#193)

8311a60
unverified

whisper : add mechanism for aborting the whisper_full() computation

d311de4

whisper.objc : add real-time processing (#97)

11bb554
unverified

whisper.objc : fix build warnings

8d1f7e9
unverified

whisper.wasm : do not block page while processing (close #86)

d0b1d9e
unverified

refactoring : more readable code

5ef0168
unverified

stream : "-kc" now enables context keeping from previous segment (#90)

28726dd
unverified

Prompt previous tokens for streaming (#163)

8ad3dbf
unverified

stream : add "audio_ctx" parameter

6adc1fe

stream : add "max_tokens" parameter

e48ba5c

stream : add "single_segment" option

a265bfa

stream : partial encoder experiments

a2015c0

whisper : add option to speed up the audio tempo by x2

bec875e

ref #22 : add "duration" option

097e96a
unverified

whisper : token-level timestamp refactoring (#49, #120)

e4f586b
unverified

main : add option for word-leve timestamps (very experimental)

b777dde

close #113 : fix struct whisper_token_data

ba9fc0a
unverified

parallel : print time of audio boundaries + fix timings

84ba527

parallel : working

d5d2d41

main : fix sampling time + add max_context parameter

9ed1355

parallel : adding tool for parallel transformer inference

cafe46d

ggml : add system info functions

7a4a7a1

whisper : add new-segment callback

83926f7

wip : experimental color coding of tokens based on probabilities

6f2f468

ref #52 : improve greedy sampling strategy

07043b9
unverified

close #32 : add comment about thread-safety of the C-style API

a3ee28a
unverified

ref #57, #62, #63 : remove unions in C-api + remove designated initializers

99f2636
unverified

ref #35 : add <stdbool.h> to whisper.h

0ab5514
unverified

ref #9 : add API documentation in whisper.h

ce2cd6d

Adding sanitizer tests

6113fbd

ref #10 : option to keep context in "stream" example

606a6bc
unverified

ref #16, #22 : add "offset" argument

35a7fcb
unverified

Improve result printing

9a513f5

Extend C-style API with full inference methods

dabc473

Initial C-style interface for whisper.cpp

c0943fb