### Welcome to "Praat: Beyond the Basics" 1) Please download the latest Version of Praat from praat.org 2) Please download the latest version of the Course Materials... http://www.savethevowels.org/praat/lsa_minicourse_materials.zip Use WiFi Network "JWMarriot_CONFERENCE" **The Room Wifi Password is LSA2017** *If you haven't yet registered, or registered after the end of the year, please email me (wstyler@umich.edu) and we'll sort everything out* --- # Praat Beyond the Basics Will Styler --- ### First, some basics --- <img class="r-stretch big" src="phonmedia/sampling_raw.png" alt="$txt"> --- ### Computers only know 0 and 1 --- ### So, how do we represent this with 0 and 1? <img class="r-stretch big" src="phonmedia/sampling_raw.png" alt="$txt"> --- ### Digitization * (Also known as "Discretization") * Instead of representing the wave as a wave, you measure the wave’s amplitude many, many times per second. * This gives you a series of points which represent the wave’s form, overall. --- ### Let's use a simpler wave <img class="r-stretch big" src="phonmedia/sampling_wave.png" alt="$txt"> --- ### Let's break it into points <img class="r-stretch big" src="phonmedia/sampling_quantized.png" alt="$txt"> --- ### And measure the amplitude at each of these points <img class="r-stretch big" src="phonmedia/sampling_measures.png" alt="$txt"> --- ### A WAV file is just a list of time/amplitude pairs * ... and fundamentally, these pairs are what Praat is reading in and doing * Any time that we're manipulating sounds, especially by formula, we're relying on this kind of math --- ### The number of samples per second is the 'sampling rate' <img class="r-stretch big" src="phonmedia/sampling_quantized.png" alt="$txt"> --- ### As the rate drops, so does the detail <img class="r-stretch big" src="phonmedia/sampling_highrate.png" alt="$txt"> --- ### As the rate drops, so does the detail <img class="r-stretch big" src="phonmedia/sampling_lowrate.png" alt="$txt"> --- ### As the rate drops, so does the detail <img class="r-stretch big" src="phonmedia/sampling_toolow.png" alt="$txt"> --- ### Until you get a full-on acoustic trainwreck <img class="r-stretch big" src="phonmedia/sampling_undersampled.png" alt="$txt"> --- ### Higher frequencies can't be captured with low sampling rates * The Highest Frequency captured is 1/2 of the Sampling rate * This is the Nyquist Theorem --- ### Sampling Rate Makes a big difference! Track: Shpongle - *Nothing is something worth doing* 44,100 Hz - <audio controls><source src="phonmedia/nothingsomething44100.wav" type="audio/wav"></audio> 22,050 Hz - <audio controls><source src="phonmedia/nothingsomething22050.wav" type="audio/wav"></audio> 11,025 Hz - <audio controls><source src="phonmedia/nothingsomething11025.wav" type="audio/wav"></audio> 6,000 Hz - <audio controls><source src="phonmedia/nothingsomething6000.wav" type="audio/wav"></audio> 3,000 Hz - <audio controls><source src="phonmedia/nothingsomething3000.wav" type="audio/wav"></audio> 1,500 Hz - <audio controls><source src="phonmedia/nothingsomething1500.wav" type="audio/wav"></audio> 800 Hz - <audio controls><source src="phonmedia/nothingsomething800.wav" type="audio/wav"></audio> --- ### We generally use 44,100 Hz as a sampling rate * This lets us capture frequencies from 0 - 22,050 Hz * Humans can hear 20-20,000 Hz (roughly) * **This sampling rate captures everything humans can hear!** * Sorry dolphins! --- ### Fundamentally, Praat is just an interface for reading, playing, and editing these samples * So, let's talk about that interface.