KO
|
EN
gitlite — search
Search
#javascript
#python
#hacktoberfest
#react
#ai
#typescript
#llm
#go
#golang
#android
#machine-learning
#rust
#deep-learning
#linux
precoss
★ 8
Open GitHub ↗
No description available.
Download README (.md)
Explore Similar Repositories
NMF
:
Implementation of Projected gradient NMF
sword-of-the-orc-king
:
Roguelike adventure game being written in Lua and LÖVE
pytorch-mobile-kit
:
PyTorch Mobile starter kit.
math240
:
MATH 240, Discrete Structures, Fall 2019 at McGill University. Handwritten lecture notes including diagrams from Rosen's "Discrete Mathematics and Its Applications."
Cloud-Hospital
:
本系统使用了大量的优化手段，如抢号（秒杀）场景的考虑、复杂计算的优化和组件化的考虑等，可适配改造于各类管理平台和购物平台等
// repository documentation
Was this content helpful?
★ 0
(0 ratings)
Select Rating:
★
★
★
★
★
Submit Feedback
Recent Feedback
×
Download README
Do you want to download the
README.md
file for
precoss
?
Download (.md)
# Description and content ## Description MATLAB code custom scripts for the manuscript as "Combining predictive coding and neural oscillations enables online syllable recognition in natural speech". Sevada Hovsepyan (sevada.hovsepyan@gmail.com) ## Content The folder named `data_construction` contains scripts that generate data for model simulations from the sound waveforms and phonemic transcriptions in the TIMIT database. Data of each sentence and list of sentences used for model simulations are stored in the `Data` folder. As an example, 10 sentences (and their simulation results) are uploaded. However the variable `Data/full_sentence_list.mat` contain the full list of sentences used in model simulations. Folders named `full_simulations` and `compressed_speech_simulations` contain scripts that run simulations on the dataset in the `Data` folder. Results are stored in subfolders in `Data`. However, the variables `perf_internal.mat` and `perf_results_compression.mat` in the folder `stat_analysis` contain performance values for all sentences for each model variant and condition. The `stat_analysis` folder also contains the scripts that were used to calculate and analyse the performance of each model variant. Furthermore, each folder also contains its own `Readme.md` file with further descriptions. ## System requirements and runtime The simulations were run on the PC desktop running Windows 7 (Intel Core i7-4790, 3.6 GHz, 16 GB RAM). Scripts should also work on Linux operating system (e.g. Ubuntu 18.04 or derivatives). Typically on the system described above simulation for one sentence takes from around 15 minutes up to 40 minutes (rarely could be more; depends on the duration of a sentence and number of syllables). ## External toolboxes and materials Sound recordings are taken from the TIMIT database. (https://catalog.ldc.upenn.edu/LDC93S1) For calculation of the auditory spectrogram the model of auditory periphery by Chi and Shamma (2005) was used https://github.com/tel/NSLtools. Finnaly, for simulations `spm_12` package is needed https://www.fil.ion.ucl.ac.uk/spm/software/spm12/ .
