soundDENA¶
soundDENA is a Python library for easily and precisely accessing various sorts of natural sounds data in bulk. It allows you to treat the NSNSD hierarchical file structure almost as though it were a queryable database.
SoundDENA:
- Provides accessor methods which handle and normalize inconsistencies in naming and file formats: input data may differ, but the output is consistent
- Associates metadata with data, so data can be selected based on a query of its metadata
- Returns data in pandas structures, which excel at concise and efficient subselection, querying, aggregation, and general wrangling
- Plays nicely with the Python scientific computing ecosystem
- Handles missing data without making a fuss
A taste of soundDENA:
In [1]: import soundDENA
In [2]: query = 'unit == "DENA" and type == "Grid" and not winter_site'
In [3]: denaSummerGrid = soundDENA.metadata.query(query)
In [4]: srcids = soundDENA.srcid.all(denaSummerGrid)
In [5]: srcids
Out[5]:
len ... tagDate
siteID ...
DENABEAR2007 2007-06-14 00:46:15 00:05:39 ... 1970-01-01
2007-06-14 01:39:11 00:05:07 ... 1970-01-01
2007-06-14 05:42:34 00:02:25 ... 1970-01-01
2007-06-14 05:45:00 00:03:39 ... 1970-01-01
... ... ... ...
DENAWFYR2006 2006-08-09 12:32:35 00:04:31 ... 1970-01-01
2006-08-09 08:15:47 00:06:39 ... 1970-01-01
2006-08-09 03:16:22 00:04:20 ... 1970-01-01
2006-08-09 02:02:57 00:02:30 ... 1970-01-01
[18301 rows x 10 columns]
# Total length of noise events, by vehicle type:
In [6]: srcids.groupby("srcID")["len"].sum()
Out[6]:
srcID
0.0 0 days 00:00:00
1.0 0 days 00:00:46
1.1 14 days 08:04:05
1.2 30 days 07:15:00
1.3 1 days 06:28:35
2.0 4 days 07:43:36
Name: len, dtype: timedelta64[ns]
Contents: