synesthesiam commited on
Commit
0cddca9
1 Parent(s): 98cc601

Moved to piper-voices

fa/fa/amir/medium/MODEL_CARD DELETED
@@ -1,15 +0,0 @@
1
- # Model card for dataculaamir (medium)
2
-
3
- * Language: fa (Farsi, Persian)
4
- * Speakers: 1
5
- * Quality: medium
6
- * Samplerate: 22,050Hz
7
-
8
- ## Dataset
9
-
10
- * URL: https://datacula.com/tts-databases
11
- * License: CC0
12
-
13
- ## Training
14
-
15
- Finetuned from U.S. English lessac voice (medium quality).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fa/fa/amir/medium/train.sh DELETED
@@ -1,17 +0,0 @@
1
- #!/usr/bin/env bash
2
- set -eo pipefail
3
-
4
- this_dir="$( cd "$( dirname "$0" )" && pwd )"
5
- repo_dir="$(realpath "${this_dir}/../../../../")"
6
-
7
- export PYTHONPATH="${repo_dir}/src/python:${PYTHONPATH}"
8
- python3 -m piper_train \
9
- --dataset-dir "${this_dir}" \
10
- --accelerator 'gpu' \
11
- --devices 1 \
12
- --batch-size 48 \
13
- --validation-split 0 \
14
- --num-test-examples 0 \
15
- --max_epochs 10000 \
16
- --checkpoint-epochs 1 \
17
- --precision 32 "$@"