Toolbox for constructing semantic speech networks from speech transcripts.
The algorithms in this toolbox create a semantic speech graph from transcribed speech. Speech transcripts are short paragraphs of largely raw, uncleaned speech-like text. For example:
'I see a man and he is wearing a jacket. He is standing in the dark against a light post. On the picture there seems to be like a park and... Or trees but in those trees there are little balls of light reflections as well. I cannot see the... Anything else because it’s very dark. But the man on the picture seems to wear a hat and he seems to have a hoodie on as well. The picture is very mysterious, which I like about it, but for me I would like to understand more about the picture.' -- Example Transcript
Below is the semantic speech graph constructed from this text.
Figure 1. Semantic Speech Graph. Nodes represents an entity mentioned by the speaker (e.g. I, man, jacket). Edges represent relations between nodes mentioned by the speaker (e.g. see, has on).
Read the full documentation here.
You can install the latest release from PyPi
pip install netts
or get the latest development version from GitHub (not stable)
pip install git+https://github.com/alan-turing-institute/netts
Netts requires the Java Runtime Environment. Instructions for downloading and installing for your operating system can be found here.
Netts also requires a few additional dependencies to work which you can download with the netts CLI that was installed by pip
netts install
The quickest way to process a transcript is with the CLI.
netts run transcript.txt outputs
where transcript.txt
is a text file containing transcribed speech and outputs
is the name of a directory to write the outputs to. Additional logging information can be found in netts_log.log
.
Netts was written by Caroline Nettekoven in collaboration with Sarah Morgan.
Netts was packaged in collaboration with Oscar Giles, Iain Stenson and Helen Duncan.