[Simon King]

You also need to uncomment the line setting the path to all the data:

# DATA=${DATA:-/Volumes/Network/courses/sp/data}

This week’s B tutorial will explain the assignment – you don’t need to start working on it until after that tutorial.

[Simon King]

The path you show is perfectly valid, although of course it seems a rather unlikely one and will probably result in a large global distance (indicating that the template and unknown are probably not very similar). The number of local distances summed up to make the global distance can indeed vary depending on the path taken through the grid.

The two sequences of feature vectors can be of differing lengths – this will generally be the case, in fact.

[Simon King]

Try forcibly quitting the VMWare host program. On a Mac this would be the Apple menu (top left of your screen), then “Force Quit…”, or via the Activity Monitor. On Windows, use the Task Manager.

[Simon King]

Work in the virtual machine. HTK is already installed. You do not need Wavesurfer because this year we are going to skip the data collection part.

[Simon King]

Don’t start the assignment until you’ve had Module 6 Tutorial B. Just read the instructions through before that tutorial.

You should have created ~/Documents/sp in the first assignment. If not, create it now.

Do not work in /Volumes/Network/courses/sp/digit_recogniser – those are the master copies which you need to make a copy of.

[Simon King]

We will be announcing details of the exam shortly – it will be a take-home timed test in the December exam diet. It was not timetabled by the central unit, and so we will need to schedule this ourselves.

[Simon King]

I always prefer making notes on paper, whether in-person or online. This is because it’s faster to write by hand than to type, and easier to draw diagrams than using electronic note-taking software. But, perhaps I am a bit old-school and others will have better advice.

The new (for 2020-21) topic videos are made with the following tools:

Waveform and spectrum plots including the animated versions: my own Python code using matplotlib, scipy, librosa

Slides, including other animations: Apple Keynote

Screen capture and video editing: Screenflow

Audio hardware: DPA headset microphone, dbx 286s mic preamp / processor, Apogee Duet audio interface

Video compression: Handbrake

[Simon King]

Can you post the complete steps needed to reproduce this, from starting festival onwards.

[Simon King]

Have you tried clicking the “restart or reconnect to it” link?

[Simon King]

There are various ways, such as these scripts, but this is beyond what is expected for the Speech Processing assignment. You should be able to find all types of errors by hand-crafting relatively short and simple test sentences.

[Simon King]

No – if a word is in the dictionary then Festival will always use that in preference to G2P. This is true even if there is no dictionary entry with the required POS: it will then ignore POS and pick another entry.

[Simon King]

The best way to understand the phone set is to look up common words that you know. See also this topic.

The determine the correct pronunciation for a word not in the dictionary, a good technique is to look up similar-sounding words and assemble a pronunciation from parts of them. For example, to work out the pronunciation of “Simonification” (a word I just made up, but a plausible word in English), I would use lex.lookup_all to ensure I am only retrieving dictionary entries and not using G2P:

festival>  (lex.lookup_all "Simon")
(("simon" nnp (((s ae) 1) ((m @ n) 0))))

festival>  (lex.lookup_all "magnify")
(("magnify" vb (((m a g) 1) ((n i) 0) ((f ae) 0)))
 ("magnify" vbp (((m a g) 1) ((n i) 0) ((f ae) 0))))

festival>  (lex.lookup_all "classification")
(("classification" nn  (((k l a s) 2) ((i) 0) ((f i) 0) ((k ei) 1) ((sh n!) 0))))

from which I would decide that “Simonification” might be pronounced as

s ae m @ n i f i k ei sh n!

I could optionally refine that by thinking about syllabification or syllable stress (perhaps “Simonification” should have stress on the second syllable so the vowel should be full, not reduced to schwa @).

I could also compare this to what the G2P model predicts:

festival>  (lex.lookup "Simonification")              
("Simonification" nil (((s ae m) 0) ((oo n) 0) ((i f) 0) ((i k) 0) ((ei sh n!) 0)))

In this case, the prediction looks reasonable.

[Simon King]

You can simply use lex.lookup_all which shows you all available entries in the lexicon for a word:

festival> (lex.lookup_all "content")
(("content" jj (((k @ n) 0) ((t e n ?) 1)))
 ("content" nn (((k oo n) 1) ((t e n ?) 0)))
 ("content" vb (((k @ n) 0) ((t e n ?) 1)))
 ("content" vbp (((k @ n) 0) ((t e n ?) 1))))

[Simon King]

I’m investigating why the exam has not yet been scheduled. It will be within that exam period.

[Simon King]

Instead of suspending, try actually rebooting (might be called restart in the VMWare menus). This is equivalent to powering the machine off and on again.

[Author]

Posts