Commit ba38ae68 authored by Olivier Galibert's avatar Olivier Galibert
Browse files

Add more documentation

parent 0e7c0237
......@@ -9,8 +9,8 @@ Contact email: <lifelong-speaker-evaluation@univ-lemans.fr>
| Registration open | September 2021 |
| Development set released | September 15, 2021 |
| Registration deadline | December 15, 2021 |
| Evaluation data available | December 15, 2021 |
| Evaluation period ends | January 15, 2022 |
| Evaluation data available | December 15, 2021 |
| Evaluation period ends | January 15, 2022 |
| System descriptions due | February 28, 2022 |
| Workshop | Possibly June 2022 |
......@@ -272,6 +272,24 @@ message_to_user = {
request
}
# Initialize the user simulation
user = UserSimulation(parameters)
`parameters` is a hash of values/booleans:
* float `max_cost_per_file`: maximum total cost allowed for a show before the user simulation refuses to answer
* float `request_collar_cost`: collar size used for computing answer costs (seconds)
* float `minimal_validation_time`: minimal speaker time to validate a speaker identification (seconds)
* float `minimal_validation_purity`: minimal speaker purity to validate a speaker identification (0-1)
* bool `question_allow_boundaries`: allow the user simulation to give boundaries of a segment (default True)
* bool `question_allow_same_intra_show`: allow the user simulation to say two points in the current show are of the same speaker (default True)
* bool `question_allow_different_intra_show`: allow the user simulation to say two points in the current show are of different speakers (default True)
* bool `question_allow_same_inter_show`: allow the user simulation to say two points in the different shows are of the same speaker (default True)
* bool `question_allow_different_inter_show`: allow the user simulation to say two points in the different shows are of different speakers (default True)
* bool `question_allow_name`: allow the user simulation to provide the name of a speaker (default True)
Every question type is allowed in __active__ mode, the restrictions
are only taken into account in __interactive__ mode.
# Send the request to the user simulation and receive the answer
human_assisted_learning, user_answer = user.validate(message_to_user)
```
......@@ -285,6 +303,15 @@ The user simulation returns two objects:
* `human_assisted_learning` a boolean, True if the system can ask more questions, False otherwise
* `user_answer`, the answer of the user simulation that is defined as follow:
# Final validation of the speakers at the end of handling a show
speakers = user.final_commit(hypothesis)
Returns a list of pairs [hypothesis speaker name, status], both strings, where status is, in order of priority:
* `known` the speaker is already mapped from previous shows
* `short` there is not enough signal to validate the speaker (speech duration < minimal_validation_time)
* `bad` the speech of the speaker is not pure enough to be validated (speech purity < minimal_validation_purity)
* `good` the speaker is validated and that name should be used when present in future shows
### Evallies API
This section describes the different objects provided as part of the evallies API.
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment