From 3d9f094e9652d4b84894c6fd4eae39a4a753b0f0 Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期二, 16 五月 2023 23:48:00 +0800
Subject: [PATCH] train
---
funasr/datasets/small_datasets/dataset.py | 13 +++++++++++--
1 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/funasr/datasets/small_datasets/dataset.py b/funasr/datasets/small_datasets/dataset.py
index 6ba8a02..e14e4f1 100644
--- a/funasr/datasets/small_datasets/dataset.py
+++ b/funasr/datasets/small_datasets/dataset.py
@@ -9,6 +9,7 @@
from typing import Collection
from typing import Dict
from typing import Mapping
+from typing import Optional
from typing import Tuple
from typing import Union
@@ -110,6 +111,8 @@
float_dtype: str = "float32",
int_dtype: str = "long",
dest_sample_rate: int = 16000,
+ speed_perturb: Union[list, tuple] = None,
+ mode: str = "train",
):
assert check_argument_types()
if len(path_name_type_list) == 0:
@@ -123,6 +126,10 @@
self.float_dtype = float_dtype
self.int_dtype = int_dtype
self.dest_sample_rate = dest_sample_rate
+ self.speed_perturb = speed_perturb
+ self.mode = mode
+ if self.speed_perturb is not None:
+ logging.info("Using speed_perturb: {}".format(speed_perturb))
self.loader_dict = {}
self.debug_info = {}
@@ -146,13 +153,15 @@
loader_type: loader_type. sound, npy, text, etc
"""
if loader_type == "sound":
- loader = SoundScpReader(path, self.dest_sample_rate, normalize=True, always_2d=False)
+ speed_perturb = self.speed_perturb if self.mode == "train" else None
+ loader = SoundScpReader(path, self.dest_sample_rate, normalize=True, always_2d=False,
+ speed_perturb=speed_perturb)
return AdapterForSoundScpReader(loader, self.float_dtype)
elif loader_type == "kaldi_ark":
loader = kaldiio.load_scp(path)
return AdapterForSoundScpReader(loader, self.float_dtype)
elif loader_type == "npy":
- return NpyScpReader()
+ return NpyScpReader(path)
elif loader_type == "text":
text_loader = {}
with open(path, "r", encoding="utf-8") as f:
--
Gitblit v1.9.1