|
Revision 1, 0.5 kB
(checked in by kmaclean, 1 year ago)
|
initial commit of Sphinx Acoustic Model creation scripts
|
| Line | |
|---|
| 1 |
To run it |
|---|
| 2 |
|
|---|
| 3 |
1. Build sphinxtrain |
|---|
| 4 |
2. Run ../SphinxTrain/scripts.pl/setup_SphinxTrain -update -task voxforge_it |
|---|
| 5 |
3. Run ./scripts.pl/make_feats.pl -ctl etc/voxforge_it_train.fileids |
|---|
| 6 |
4. Run ./scripts.pl/RunAll.pl |
|---|
| 7 |
5. Get the model |
|---|
| 8 |
|
|---|
| 9 |
To update it with new prompts: |
|---|
| 10 |
|
|---|
| 11 |
1. Update wav files |
|---|
| 12 |
2. Update transcription |
|---|
| 13 |
3. Update fileids and dictionary |
|---|
| 14 |
4. Run make_feats |
|---|
| 15 |
5. Run RunAll.pl as above |
|---|
| 16 |
|
|---|
| 17 |
|
|---|
| 18 |
To decode sound with model: |
|---|
| 19 |
|
|---|
| 20 |
1. Put sound it wav |
|---|
| 21 |
2. Update test.fileids |
|---|
| 22 |
3. Run ./scripts.pl/make_feats.pl -ctl etc/voxforge_it_test.fileids |
|---|
| 23 |
4. Run ./scripts_pl/decode/slave.pl |
|---|
