# p_discolor_auditory_feedback Acoustic parameters of P. discolor vocalisations, auditory brainstem responses, and data analysis programs for Lattenkamp et al., 2021, “The vocal development of the pale spear-nosed bat is dependent on auditory feedback”. ## Vocalisations The `vocalisations` directory provides data in four CSV files: * `pup_calls.csv`, `adult_calls.csv`: Acoustic call parameters of the vocalisations recorded from six bats, three of which were deafened after the first recording session. In `pup_calls.csv`, these bats were up to half a year old; in `adult_calls.csv`, the same individuals were 2 years old. The juvenile recordings were acquired with pairs of bats: In each session, one (deafened or normal-hearing) juvenile was paired with its (always normal-hearing) mother. The adult recordings were acquired from groups of two or three bats such that within each recording session, all were either deafened or of normal hearing. `pup_calls.csv` includes the vocalisations from the juvenile bats’ respective mothers which were placed in the recording chamber with their offspring in some of the sessions. These data have not been analysed further by the authors. * `pup_sessions.csv`, `adult_sessions.csv`: Metadata for the recording sessions. ### Acoustic call parameters Each row of `*_calls.csv` corresponds to one extracted vocalisation from one bat. The columns are: * `session_id`: ID of the recording session in which this vocalisation occurred, for lookup in the corresponding `*_sessions.csv` file. * `call_id`: Sequence number of the vocalisation within the recording session. The pairs (`session_id`, `call_id`) are unique within each CSV file. * `start_sample`: Sample index in the microphone buffer where the extracted vocalisation started. * `start_time`: Absolute time in Central European (Summer) Time at which the extracted vocalisation started. Can be calculated from `start_sample` and the metadata in `*_sessions.csv`, based on the sampling rate of 192 kHz. * `calling_bat` (`pup_calls.csv` only): Identifier of the animal which emitted the vocalisation. This can be determined by comparing levels in the microphone array in the recording chamber, as the position of the bats within the chamber was known and fixed. This column is missing from `adult_calls.csv`, where the recording setup did not facilitate this analysis. * `other_bat` (`pup_calls.csv` only): Identifier of the animal which was in the recording chamber with the calling bat, but did not emit the vocalisation. Missing in `adult_calls.csv`, where this information is unknown. * `calling_bat_deaf` (`pup_calls.csv` only): 1 if the calling bat was deafened, 0 otherwise. Note that in `pup_calls.csv`, the first session of each juvenile–mother pairing was recorded before deafening, yet `calling_bat_deaf` is 1 even in those sessions. * `group` (`adult_calls.csv` only): The string “hearing” or “deaf”, reflecting the group that all individuals in the recording session belonged to. * `before_deafening` (`pup_calls.csv` only): 1 if this vocalisation occurred in the first session of a juvenile (before any deafening), 0 otherwise. * `calling_bat_mother` (`pup_calls.csv` only): 1 if the calling bat was a mother, 0 otherwise. * `level_difference` (`pup_calls.csv` only): Difference between the root-mean-square vocalisation levels of the microphones closer to one bat vs. the microphones closer to the other bat. Used to determine `calling_bat` and `other_bat` in sessions with juveniles. * `call_rms` (in dB): Root-mean-square sound pressure of the vocalisation at the microphone were it was loudest, relative to an arbitrary reference which was constant across sessions in `pup_calls.csv` and `adult_calls.csv`, but not the same in both. * `call_duration` (in s): Duration of the vocalisation. * `mean_aperiodicity`: Aperiodicity value extracted with the YIN algorithm for fundamental frequency estimation. * `f0_mean`, `f0_min`, `f0_max`, `f0_start`, `f0_end`, `f0_slope` (in Hz): Mean, minimum, maximum, initial and final fundamental frequencies of the vocalisation, respectively, as estimated with the YIN algorithm. * `spectral_centroid` (in Hz): The center of mass of the frequency spectrum of the vocalisation. ### Session files `pup_sessions.csv` contains the session IDs (`session_id`), start times in Central European (Summer) Time (`start_time`), the IDs of the two individuals which were placed in the recording chamber together in each session (`animal1` and `animal2`), and `before_deafening` indicating whether a recording was the first of a juvenile. `adult_sessions.csv` is structured similarly, but replaces the two `animalX` columns with `group`, which is either `deaf` or `hearing`, and does not contain `before_deafening`. ### Animals The individuals referred to in the data files were: * `b2`, born on 2017/01/26, never deafened. * `b3`, born on 2017/01/29, deafened after the first recording session. * `b4`, born on 2017/01/30, never deafened. * `b5`, born on 2017/02/02, deafened after the first recording session. * `b7`, born on 2017/02/08, never deafened. * `b8`, born on 2017/02/08, deafened after the first recording session. * `mh`, `mt`, `mb`, `my` and `mp`, the mothers of these six bats. ### Scripts `scripts` contains the Python script which was used to create the figures and tables included in the publication, and the R script which was used to conduct the statistical analysis. `scripts_for_reference` is of documentary nature and contains the MATLAB and Python code which was used to extract the acoustic parameters and generate the CSV files from the raw microphone data (which is not included in full in this repository for reasons of space, but see the section below). The YIN implementation used by some of the MATLAB scripts is available from http://audition.ens.fr/adc/sw/yin.zip (de Cheveigné & Kawahara, JASA 111:1917:1930, 2002, doi:10.1121/1.1458024). ### Exemplary raw data A sample of raw data files is included in this dataset. Specifically, 18 exemplary recording sessions (including data from all six juvenile bats) are provided as 2-channel, 192 kHz, 24-bit PCM audio data. For more efficient storage, these have been compressed using a lossless codec and are provided in FLAC containers in the `vocalisations/pup_recordings` directory. The filename schema is `pup_session_SSS_AABB.flac`, where `SSS` corresponds to the `session_id` and `AA`/`BB` to the `animal1`/`animal2` fields of `pup_sessions.csv`. The first channel in each file was recorded with a microphone closer to the `AA` bat, while the microphone associated with the second channel was closer to the `BB` bat. ## Auditory brainstem responses The `abr_analysis_results.mat` file in the `ABR` directory is a MATLAB data file with three variables: * `dates`: Recording dates of each ABR measurement sessions, formatted as YYYYMMDD strings. * `bats`: Identity of the animal in each ABR measurement session. Note that the first (earliest) recording for each bat was acquired before any deafening. * `allCONFI`: A structure array, indexed on ABR measurement session number and containing three matrices, namely - `frequencies`, the frequencies of the stimulation tone pips in Hz, - `levels`, the tested sound levels in dB, and - `confidences`, a confidence value for whether the tone pip specified by frequency and level evoked an auditory brainstem response, based on a bootstrap analysis. * `resultTHRES`: A structure array, indexed on ABR measurement session number and containing two matrices, namely - `uniqueFrequencies`, the unique frequencies of the stimulation tone pips in Hz, and - `bootstrapThresholds`, the lowest sound level for the corresponding frequency for which an auditory brainstem response was obtained with a confidence value of 0.95 or greater, or NaN if there is no such level. `abr_data_analysis.m`, provided for reference, is the MATLAB script by which these data were extracted from the raw ABR measurements. `abr_data_plot.m` was used to generate the supplementary ABR audiometry figure.