FreeDV
/
phasenn
mirror of https://github.com/drowe67/phasenn.git
1
0
Fork 0
Code Issues Releases Wiki Activity
71 Commits
1 Branch
0 Tags
252 KiB
Commit Graph

71 Commits (master)

Author SHA1 Message Date
David 6171230b92 updated README 2020-11-07 17:01:52 +10:30
David 123213d16d converted phasenn_test8 to tf.keras, this script has a custom loss function 2020-11-07 16:50:48 +10:30
David 479a2580fb updated README, phasenn_test9 ported ed to tf.keras 2020-11-07 16:10:16 +10:30
David cfb446155d moved amplitude specific simulations to a new repo 2019-12-23 12:58:08 +10:30
David 041bb29ec3 first pass at eband->{Am}, getting quite reasonable SD results and test plots for 10-14 length vectors 2019-12-22 13:06:43 +10:30
David 628dca7638 restored to original NN model 2019-12-21 14:04:28 +10:30
David d2039fd4b7 cleaned up and example cmd lines 2019-12-21 13:56:12 +10:30
David b65abfe150 added per frame linear const est 2019-12-20 13:25:33 +10:30
David 9076bd73fb building up script to explore interpolation/decimation of rate K vectors using NNs 2019-12-18 05:58:06 +10:30
David 900b286bcc adding results, work in progress 2019-12-15 18:34:21 +10:30
David 2ff2166344 comparing with conventionally decoded vectors, some issues as conventional has benign errors at end points 2019-12-15 18:34:06 +10:30
David 5c4367d8a1 Revert "quite reaonable results on visible inspection of a few frames" 
This reverts commit fc79915bab.
 2019-12-13 17:36:59 +10:30
David fc79915bab quite reaonable results on visible inspection of a few frames 2019-12-13 17:36:47 +10:30
David d372e37903 quite reaonable results on visible inspection of a few frames 2019-12-13 17:36:32 +10:30
David 76ffb675e4 first pass at newamp1_train 2019-12-13 16:36:41 +10:30
David aa2087ebdd add phase0 model to compare 2019-12-13 13:32:33 +10:30
David 712b1d5198 using voiced/unvoiced model in synth 2019-12-13 09:08:34 +10:30
David a2832ecb6a training on just high energy V frames, didn't improve randomness of UV speech 2019-12-13 08:15:34 +10:30
David 837de6367c tests with synthetic speech to attempt to track down UV issueson real speech 2019-12-11 05:32:54 +10:30
David c1484a76c1 fixed custom loss function - good matches and even overfitting - but still probs with UV 2019-12-09 20:22:05 +10:30
David 5c19633779 adding vanilla codec2 as a comparison 2019-12-08 17:46:27 +10:30
David 1b5b854c61 phasenn_train.py used for initial synthesis 2019-12-08 17:21:40 +10:30
David 67adf33ebe building up synthesis tools 2019-12-08 17:18:04 +10:30
David 62c5474cf6 first pass at synth script to put everything back together 2019-12-08 12:55:17 +10:30
David 03722ac424 simulated V/UV speech, used to test phasenn_train.py, with this signal, UV strain sup well (nice and random) 2019-12-08 12:54:40 +10:30
David de7b7ea394 amplitude waiting doing OK for a few test frames of voiced speech, UV too clicky, suspect random phases being averaged, 22 minute train_8k 2019-12-07 14:28:57 +10:30
David 0b34b92906 first pass phase training, no t overly impressed by results, but basic frameowkr doing sensible things, e.g. overtraining on small, all_8k database 2019-12-07 11:58:21 +10:30
David fd14956b60 first pass with all_8k, converges _too_ well, firs UV phase perfectly 2019-12-07 10:37:26 +10:30
David 6c8a2052ff added better visualisation of phase and amplitude together in PNGs 2019-12-06 12:29:12 +10:30
David f0ee4a7684 added CLI, ability to work with internal and external n0 removal 2019-12-06 09:10:52 +10:30
David 329bbf710f changed cost function for est_n0, getting reasonable results for simulated speech and real speech vectors make some sort of sense (hard to tell with phase spectra) 2019-12-04 16:21:54 +10:30
David 3a70ff4677 Merge branch 'master' of github.com:drowe67/phasenn 2019-12-04 14:26:43 +10:30
David 4f8c7ea4fb added some tests for est_n0, working OK for basic impulse case 2019-12-04 14:26:24 +10:30
David 9ab6a221dc debugging plot_n0.py/est_n0 C code 2019-12-03 07:07:59 +10:30
David 56abe1afd2 utility to plot n0 for Codec 2 model files 2019-12-01 18:41:14 +10:30
drowe67 0a39182a84
Update README.md 2019-12-01 11:08:43 +10:30
drowe67 12ccecab68
Update README.md 2019-12-01 11:07:45 +10:30
drowe67 57ccb4c122
Update README.md 2019-12-01 11:06:23 +10:30
drowe67 900fdbfa3f
Update README.md 2019-12-01 11:03:14 +10:30
drowe67 b3a25900a1
Update README.md 2019-12-01 10:55:10 +10:30
drowe67 290323bf81
Update README.md 2019-12-01 10:27:46 +10:30
drowe67 0d6129d500
Update README.md 2019-12-01 10:25:23 +10:30
David 0b605d3af5 swapped example names 2019-12-01 10:23:42 +10:30
David cb8cb985ac example images 2019-12-01 10:05:23 +10:30
David 3ef98f55a7 codec 2 model record I/O test working OK 2019-12-01 09:50:58 +10:30
David 63a1db65e6 still now working 2019-11-30 13:53:18 +10:30
David 7b448657bc added V/UV model and actually doing a reasonable job 2019-11-29 18:21:10 +10:30
David 39827bfdf1 merge 2019-11-29 18:21:05 +10:30
David 3173bb3961 building up V/UV model 2019-11-24 10:22:48 +10:30
David bca28cc34a docu 2019-11-24 10:06:52 +10:30