taco/microWakeWord-Trainer-Nvidia-Docker
1
0
Fork 0
mirror of https://github.com/TaterTotterson/microWakeWord-Trainer-Nvidia-Docker.git synced 2026-06-12 20:10:19 -06:00
Code Issues Packages Projects Releases Wiki Activity

blackwell/wham & chim datasets

Browse Source
This commit is contained in:
MasterPhooey
2026-03-09 19:48:35 -05:00
parent 4c4750a7bd
commit 94903783cb
7 changed files with 517 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
cli/wake_word_sample_augmenter
View File

@@ -23,6 +23,8 @@ parser.add_argument("--personal-output-dir", type=str, help="Personal features o
parser.add_argument("--mit-rirs-16k-dir", type=str, help="MIT RIR input directory. Default: <data-dir>/training_datasets/mit_rirs_16k", required=False)
parser.add_argument("--fma-16k-dir", type=str, help="FMA input directory. Default: <data-dir>/training_datasets/fma_16k", required=False)
parser.add_argument("--audioset-16k-dir", type=str, help="Audioset input directory. Default: <data-dir>/training_datasets/audioset_16k", required=False)
parser.add_argument("--wham-16k-dir", type=str, help="WHAM input directory. Default: <data-dir>/training_datasets/wham_16k", required=False)
parser.add_argument("--chime-16k-dir", type=str, help="CHiME input directory. Default: <data-dir>/training_datasets/chime_16k", required=False)
try:
args = parser.parse_args()
@@ -71,6 +73,16 @@ if not args.audioset_16k_dir:
else:
args.audioset_16k_dir = os.path.realpath(args.audioset_16k_dir)
if not args.wham_16k_dir:
args.wham_16k_dir = os.path.join(args.data_dir, "training_datasets", "wham_16k")
else:
args.wham_16k_dir = os.path.realpath(args.wham_16k_dir)
if not args.chime_16k_dir:
args.chime_16k_dir = os.path.join(args.data_dir, "training_datasets", "chime_16k")
else:
args.chime_16k_dir = os.path.realpath(args.chime_16k_dir)
def validate_directories(paths):
for path in paths:
if not os.path.exists(path):
@@ -78,7 +90,15 @@ def validate_directories(paths):
return False
return True
required = [work_dir, args.input_dir, args.mit_rirs_16k_dir, args.fma_16k_dir, args.audioset_16k_dir]
required = [
work_dir,
args.input_dir,
args.mit_rirs_16k_dir,
args.wham_16k_dir,
args.chime_16k_dir,
args.fma_16k_dir,
args.audioset_16k_dir,
]
if not validate_directories(required):
parser.print_help()
sys.exit(1)
@@ -117,7 +137,12 @@ from microwakeword.audio.spectrograms import SpectrogramGeneration
START_TIME = datetime.now(timezone.utc).replace(microsecond=0)
impulse_paths = [args.mit_rirs_16k_dir]
background_paths = [args.fma_16k_dir, args.audioset_16k_dir]
background_paths = [
args.wham_16k_dir,
args.chime_16k_dir,
args.fma_16k_dir,
args.audioset_16k_dir,
]
augmenter = Augmentation(
augmentation_duration_s=3.2,
@@ -245,4 +270,4 @@ END_TIME = datetime.now(timezone.utc).replace(microsecond=0)
et = END_TIME - START_TIME
print(f"\n{'=' * 80}")
print(f"{'Augmentation completed.':>50s} Elapsed time: {et!s}")
print(f"{'=' * 80}\n")
print(f"{'=' * 80}\n")