AKUSTYK IS FREE

Akustyk is distributed under General Public Lincense (as is its parent program, Praat). It is therefore free of charge.

EXTENSIBLE ARCHITECTURE

Akustyk has an extensible architecture. Thanks to contributions from Akustyk users, vowel systems of other languages and dialects are being added. Please, check the current list of supported languages and dialects. Users will be able to easily switch between vowel systems and combine them all in one database, if needed.

ACCURATE MEASUREMENT OF ACOUSTIC PARAMETERS

Akustyk helps achieve the highest level of precision, consistency, and automation in the analysis of

• formants (in Hz and Bark)
• formant bandwidths
• formant amplitudes
• voicing (5 different parameters)
• duration
• intensity

INTERVAL ANALYSIS

Akustyk offers an automated interval analysis of formants and bandwidths. With one mouse click, Akustyk dynamically segments the vowel nucleus into frames of about 10ms and returns formant and bandwidth values at each frame. This is ideal for quantifying formant transitions, preceding and following environments, diphthongs, and other time-varying acoustic parameters.

QUANTIFICATION OF FORMANT TRAJECTORIES

Akustyk offeres several robust algorithms to quantify time-varying frequency components along formant trajectories.

COMPREHENSIVE AND TIME-SAVING DATA CODING

With one simple form, you can easily encode over 30 parameters including talker ID, sex, preceding and following environment, stress, as well as two user-defined categories of five options each.

FAST AND EASY DATA LOGGING

Akustyk enables you to automate the data logging process. With one mouse click, over 100 different variables are written to a standard spreadsheet for each analysis point.

FORMANT NORMALIZATION

Akustyk offers four of the most robust normalization algorithms and leaves it up to the researcher to decide which one works best in their analysis. In addition, Akustyk provides an on-the-fly comparison across data sets normalized by means of those different methods.
Akustyk includes the following normalization schemes:

• Lobanov (1971) - it standardizes the mean and standard deviation for each talker's vowel system (in Hz).
• Nearey (1977) - a log-mean normalization technique (in Hz). It is currently used in William Labov's Plotnik.
• Nordström and Lindblom (1975) - an intrinsic formant normalization technique based on an estimate of vocal tract differences between talkers (in Hz).
• Syrdal and Gopal (1986) - intrinsic normalization model based on psychoacoustic properties of speech perception (in Bark).

STATISTICAL PROCEDURES

Akustyk offers two commonly used statistical procedures to establish differences between formants across different talker populations (in Hertz and in Bark).

• T-test - use a simple form to tell Akustyk what vowels you want to compare and click OK. Your analysis is ready instantaneously.
• discriminant analysis - plot your vowels in the discriminant plane with this point-and-click procedure.

CREATE PLOTS AND GRAPHS WITH EASE

Akustyk allows you to create high-quality vowel plots for your publications.

• choose from over 10 different plotting modes to instantly plot vowel systems of different speaker populations.
• mix normalized and non-normalized data, compare individual vowels, contrast up to 8 vowel systems, use reversed and non-reversed axes, and much more.
• all of the graphs offer a full set of IPA symbols.
• dynamically create graphical representation of FFT and LPC spectra with a single mouse click
• all graphs can also be saved in EPS format

BATCH CONVERTER

Resample and convert your audio files with ease. Akustyk can convert to and from

• MS .wav
• Apple .aiff
• Sun .su
• Kay Elemetrics (CSL) .ns