Hi,
I am converting speech to text, and here is what was done:
$ deepspeech --model deepspeech-0.7.4-models.pbmm --scorer deepspeech-0.7.4-models.scorer --audio audio/gradeMeetingRecording.wav
Loading model from file deepspeech-0.7.4-models.pbmm
TensorFlow: v1.15.0-24-gceb46aae58
DeepSpeech: v0.7.4-0-gfcd9563f
2020-07-09 23:16:26.746322: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Loaded model in 0.0152s.
Loading scorer from files deepspeech-0.7.4-models.scorer
Loaded scorer in 0.0043s.
Warning: original sample rate (44100) is different than 16000hz. Resampling might produce erratic speech recognition.
Running inference.
now all these and he recorded i would like to tallahassee he confessed a man meteorite hit now is being bored i like to lie to me in the sinking inaccessibility to do see just some facts and of assertion to tottapotomoi know can you give me a moment before beaurepaire procrastination or supperstition sensibilities considerations proved secession torsoes of ale together caramanians quarter and position wistmore in the short beard of tongues was the chief de using pizenin other resources on pegs sarangani had questioned with one to one of matching questions on the quiz were coy and patient on boosting and one or two one mathieson antwerp in addition people also who were together in groups or who sees or who moustier searching for hankes this is a consequence of osiris desirous as fair and cheers pasoeroean cheating on this quarter when i was aware of what was outrageous and what was not i invested wooden recorded time too and used all my free time from fishing other science earlier decided frescoes without being distraction and with a nice potatoes beamy shore was lackington having done everything honest without using any one proposes you might hardly let my sakes not efficient for that nursing is not much a year but information does not linotypers for the supervisor gather fast are well above the average and even though she was for is but to be class and the other nursing what do i have citations the grass i send an equal napolitano study for the exams and in nursing and psychokinesis i foresight the average and i sitooation that was considered was not nicolette average and mongolian this sea is stated very well for those cases where a film because these cases are designed to pass students who who used on approved resources it is evident that to wisenham for horses with the open book to pass other teams that operation time two hours apostolic wariest make sensation boatheader thing in memory and are able to answer for his tomcattin then using book to search for answers and wooden therefore the busily siwash stipends for networks that ortter on authorisation that are not surprised it falernian plesiosaurus asiatics paracentesis the contaminating brother exams colliquative sensationalistic waistcoats classmates also ottomanisation praiseworthiness the seisachtheia jeopardise sailors oceanographers salaciousness maranatha for satirical gesticulation spiritualising a seven acidulating retirements attestations hickories caterers to help me exaltations dimensions saintesses statistics mataswintha basilicas lancastrians conciliatoriness heartrending accentuate charnesseuil challenging generally on the department chirosciphia of foreseeing from the nishinams in the past we deracinate college seasickness penitence were forehead to caparisoned secretaries ratiocination is not the eeriness i will say this for overheated back to mister tickletoeteaser's passanitello licentiousness is tintabel yet i did make
Inference took 631.990s for 1283.243s audio file.
The results are horrible. Any suggestions on what is wrong?
