Whisper does it really use my GPU , after a series of .wav getting transcribed suddenly it gets slow and get back again

[praveenRI007]

im trying to run whisper AI in my local , im feeding audio chunks continously to the model for transcribing , but thing is after a series of wav files that are transcribed suddenly gets long delay for one file and gets back again and repeats the same .

GPU specs : nvidia 1050Ti 4GB
RAM : 24GB
Processor : Intel(R) Core(TM) i5-8300H CPU @ 2.30GHz 2.30 GHz

Note: all .wav are of almost same size

below logs from my app for reference :

checkpoint = torch.load(fp, map_location=device)
model loaded !
recording started ...
Saved segment_1.wav
Saved segment_2.wav
Saved segment_3.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_4.wav
Saved segment_5.wav
Saved segment_6.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_7.wav
Saved segment_8.wav
Saved segment_9.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_1.wav: 24.367998838424683
transcribed: segment_1.wav
o/p:
[]
time taken whisper segment_2.wav: 1.2639999389648438
transcribed: segment_2.wav
o/p:
[]
time taken whisper segment_3.wav: 1.242997169494629
transcribed: segment_3.wav
o/p:
Saved segment_10.wav
time taken whisper segment_4.wav: 1.1330022811889648
transcribed: segment_4.wav
o/p:
['A refund of retail balance and metro balance']
time taken whisper segment_5.wav: 1.4770002365112305
transcribed: segment_5.wav
o/p:
['A refund of retail balance and metro balance']
Saved segment_11.wav
time taken whisper segment_6.wav: 1.4310331344604492
transcribed: segment_6.wav
o/p:
time taken whisper segment_7.wav: 1.09596848487854
transcribed: segment_7.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.']
time taken whisper segment_8.wav: 1.3529982566833496
transcribed: segment_8.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.']
Saved segment_12.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_9.wav: 1.7170016765594482
transcribed: segment_9.wav
o/p:
time taken whisper segment_10.wav: 1.0889992713928223
transcribed: segment_10.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.']
Saved segment_13.wav
time taken whisper segment_11.wav: 1.253998041152954
transcribed: segment_11.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.']
time taken whisper segment_12.wav: 1.5629985332489014
transcribed: segment_12.wav
o/p:
Saved segment_14.wav
time taken whisper segment_13.wav: 1.049997329711914
transcribed: segment_13.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.']
time taken whisper segment_14.wav: 1.2440013885498047
transcribed: segment_14.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.']
Saved segment_15.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_15.wav: 1.6109538078308105
transcribed: segment_15.wav
o/p:
Saved segment_16.wav
time taken whisper segment_16.wav: 1.2040376663208008
transcribed: segment_16.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.', 'In case card is lost then balance will be transferred on a replacement card in 7 working days.']
Saved segment_17.wav
time taken whisper segment_17.wav: 1.3010025024414062
transcribed: segment_17.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.', 'In case card is lost then balance will be transferred on a replacement card in 7 working days.']
Saved segment_18.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_18.wav: 1.5500001907348633
transcribed: segment_18.wav
o/p:
Saved segment_19.wav
time taken whisper segment_19.wav: 1.3730010986328125
transcribed: segment_19.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.', 'In case card is lost then balance will be transferred on a replacement card in 7 working days.', 'Post this a balance transfer from a retail balance to transit balance.']
Saved segment_20.wav
time taken whisper segment_20.wav: 1.5129919052124023
transcribed: segment_20.wavo/p:

['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.', 'In case card is lost then balance will be transferred on a replacement card in 7 working days.', 'Post this a balance transfer from a retail balance to transit balance.']
Saved segment_21.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_21.wav: 1.3510043621063232
transcribed: segment_21.wav
o/p:
Saved segment_22.wav
time taken whisper segment_22.wav: 1.2989652156829834
transcribed: segment_22.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.', 'In case card is lost then balance will be transferred on a replacement card in 7 working days.', 'Post this a balance transfer from a retail balance to transit balance.', 'Would take place to maintain threshold balance at transit or metro.']
Saved segment_23.wav
time taken whisper segment_23.wav: 1.3910000324249268
transcribed: segment_23.wav
o/p:
['A refund of retail balance and metro balance', 'The card will be processed in 7 working days by the nodal branch of the WISBA.', 'The bank also reserved the right to debit the card account for on offline transactions which were not recorded at the time.', 'Surrounding card. On card closure, the refund request shall be processed by Nodal SBA branch on.', 'In case card is lost then balance will be transferred on a replacement card in 7 working days.', 'Post this a balance transfer from a retail balance to transit balance.', 'Would take place to maintain threshold balance at transit or metro.']
Saved segment_24.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_24.wav: 1.7309966087341309
transcribed: segment_24.wav
o/p:
Saved segment_25.wav
recording ended !
C:\Users\prave\AppData\Local\Programs\Python\Python312\Lib\site-packages\whisper_init_.py:146: FutureWarning: You are using torch.load with weights_only=False (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for weights_only will be flipped to True. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via torch.serialization.add_safe_globals. We recommend you start setting weights_only=True for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
checkpoint = torch.load(fp, map_location=device)
model loaded !
recording started ...
Saved segment_1.wav
Saved segment_2.wav
time taken whisper segment_1.wav: 3.916999101638794
transcribed: segment_1.wavo/p:

[]
time taken whisper segment_2.wav: 1.353999137878418
transcribed: segment_2.wav
o/p:
[]
Saved segment_3.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_3.wav: 1.4659662246704102
transcribed: segment_3.wav
o/p:
Saved segment_4.wav
time taken whisper segment_4.wav: 1.2789990901947021
transcribed: segment_4.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.']
Saved segment_5.wav
time taken whisper segment_5.wav: 1.2969989776611328
transcribed: segment_5.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.']
Saved segment_6.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_6.wav: 1.3370330333709717
transcribed: segment_6.wav
o/p:
Saved segment_7.wav
time taken whisper segment_7.wav: 1.2120022773742676
transcribed: segment_7.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card']
Saved segment_8.wav
time taken whisper segment_8.wav: 1.3780007362365723
transcribed: segment_8.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card']
Saved segment_9.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_9.wav: 1.642998456954956
transcribed: segment_9.wav
o/p:
Saved segment_10.wav
time taken whisper segment_10.wav: 1.266000509262085
transcribed: segment_10.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits']
Saved segment_11.wav
time taken whisper segment_11.wav: 1.4909656047821045
transcribed: segment_11.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits']
Saved segment_12.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_13.wav
Saved segment_14.wav
Saved segment_15.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_12.wav: 12.123023509979248
transcribed: segment_12.wav
o/p:
Saved segment_16.wav
time taken whisper segment_13.wav: 1.2029781341552734
transcribed: segment_13.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits', 'erm often']
time taken whisper segment_14.wav: 1.4110031127929688
transcribed: segment_14.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits', 'erm often']
Saved segment_17.wav
time taken whisper segment_15.wav: 2.0399959087371826
transcribed: segment_15.wav
o/p:
time taken whisper segment_16.wav: 1.245002269744873
transcribed: segment_16.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits', 'erm often', "and the label for all changes in the cut. BANQ will communicate the unending terms by hosting the same on BANQ's"]
Saved segment_18.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_17.wav: 1.4889976978302002
transcribed: segment_17.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits', 'erm often', "and the label for all changes in the cut. BANQ will communicate the unending terms by hosting the same on BANQ's"]
time taken whisper segment_18.wav: 1.3430016040802002
transcribed: segment_18.wav
o/p:
Saved segment_19.wav
Saved segment_20.wav
Saved segment_21.wav
recording ended ... next chunk begins !
recording started ...
time taken whisper segment_19.wav: 7.967999696731567
transcribed: segment_19.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits', 'erm often', "and the label for all changes in the cut. BANQ will communicate the unending terms by hosting the same on BANQ's", 'or in any other manner as decided by a ban from time to time.']
Saved segment_22.wav
Saved segment_23.wav
Saved segment_24.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_25.wav
Saved segment_26.wav
Saved segment_27.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_28.wav
Saved segment_29.wav
Saved segment_30.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_31.wav
Saved segment_32.wav
Saved segment_33.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_34.wav
Saved segment_35.wav
time taken whisper segment_20.wav: 43.0679669380188
transcribed: segment_20.wav
o/p:
['The bank reserves the right to debit the card account for offline transactions.', 'not recorded at the time of loss of card', 'Shall have the absolute discretion to amend or delete or supply any of the terms features and benefits', 'erm often', "and the label for all changes in the cut. BANQ will communicate the unending terms by hosting the same on BANQ's", 'or in any other manner as decided by a ban from time to time.']
Saved segment_36.wav
recording ended ... next chunk begins !
recording started ...
Saved segment_37.wav
Process Process-3:
recording started ...

Process finished with exit code -1