Audio Analysis

1
# Required key in each data item
2
{
3
    "audio_filepath": "/path/to/audio.wav"
4
}

1
from nemo_curator.stages.audio.common import GetAudioDurationStage
2
3
# Calculate audio duration for each file
4
duration_stage = GetAudioDurationStage(
5
    audio_filepath_key="audio_filepath",
6
    duration_key="duration"
7
)

1
from nemo_curator.stages.audio.common import PreserveByValueStage
2
3
# Keep samples between 1 and 15 seconds
4
min_duration_filter = PreserveByValueStage(
5
    input_value_key="duration",
6
    target_value=1.0,
7
    operator="ge"
8
)
9
max_duration_filter = PreserveByValueStage(
10
    input_value_key="duration",
11
    target_value=15.0,
12
    operator="le"
13
)

1
import soundfile as sf
2
3
# Check if file is readable
4
try:
5
    info = sf.info("audio_file.wav")
6
    print(f"Duration: {info.duration}s, Sample rate: {info.samplerate}Hz")
7
except Exception as e:
8
    print(f"File validation failed: {e}")

1
from nemo_curator.pipeline import Pipeline
2
from nemo_curator.stages.audio.common import GetAudioDurationStage, PreserveByValueStage
3
from nemo_curator.stages.audio.inference.asr_nemo import InferenceAsrNemoStage
4
5
# Create analysis pipeline
6
pipeline = Pipeline(name="audio_analysis")
7
8
# 1. Calculate duration (handles format validation automatically)
9
pipeline.add_stage(GetAudioDurationStage(
10
    audio_filepath_key="audio_filepath",
11
    duration_key="duration"
12
))
13
14
# 2. Filter by duration (removes corrupted files with duration = -1.0)
15
pipeline.add_stage(PreserveByValueStage(
16
    input_value_key="duration",
17
    target_value=1.0,
18
    operator="ge"  # >= 1 second
19
))
20
21
pipeline.add_stage(PreserveByValueStage(
22
    input_value_key="duration", 
23
    target_value=15.0,
24
    operator="le"  # <= 15 seconds
25
))
26
27
# 3. Continue with ASR inference on validated files
28
pipeline.add_stage(InferenceAsrNemoStage(
29
    model_name="nvidia/stt_en_fastconformer_hybrid_large_pc"
30
))