audio_utils improvements #21998
sgugger
merged
39 commits into
huggingface:main
from
hollance:audio_utils
ArthurZucker
approved these changes
on 2023-03-27
silly change to allow making a PR
ff3f4079
clean up doc comments
1050de58
simplify hertz_to_mel and mel_to_hertz
fc590c77
fixup
5c27568f
clean up power_to_db
60862246
also add amplitude_to_db
dc98cc5a
move functions
e580d27f
clean up mel_filter_bank
97813931
fixup
37a153fb
credit librosa & torchaudio authors
a7bcbd15
add unit tests
c86e6bbd
tests for power_to_db and amplitude_to_db
8c846fcd
add mel_filter_bank tests
95e33613
rewrite STFT
4e690917
add convenience spectrogram function
d55a1e8c
missing transpose
a0a67a2e
fewer transposes
94bde14e
add integration test to M-CTC-T
94590b85
frame length can be either window or FFT length
ec247574
rewrite stft API
bcb6c79e
add preemphasis coefficient
41b8501f
move argument
e8663b0f
add log option to spectrogram
4eedc5cf
replace M-CTC-T feature extractor
71656b85
fix api thing
a3120c7b
replace whisper STFT
99c1ce6e
replace whisper mel filters
f8966507
replace tvlt's stft
890fa72b
allow alternate window names
1b2026e5
replace speecht5 stft
a4680c99
fixup
067c87c8
fix integration tests
22608a02
fix doc comments
534c07a4
remove manual FFT length calculation
31f30fdb
fix docs
d3144c5b
go away, deprecation warnings
7151ba0f
combine everything into spectrogram function
dd1046b8
add deprecated functions back
95827209
fixup
b3eba991
sgugger
approved these changes
on 2023-05-08
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub