From 60c8f036e0fd3e29a0334c577f5f7d91f8b01982 Mon Sep 17 00:00:00 2001
From: 仁迷 <haoneng.lhn@alibaba-inc.com>
Date: 星期五, 17 三月 2023 20:09:02 +0800
Subject: [PATCH] update audio type check
---
funasr/datasets/iterable_dataset.py | 18 ++++++------------
1 files changed, 6 insertions(+), 12 deletions(-)
diff --git a/funasr/datasets/iterable_dataset.py b/funasr/datasets/iterable_dataset.py
index c8c51d4..4b2fb1a 100644
--- a/funasr/datasets/iterable_dataset.py
+++ b/funasr/datasets/iterable_dataset.py
@@ -228,13 +228,9 @@
name = self.path_name_type_list[i][1]
_type = self.path_name_type_list[i][2]
if _type == "sound":
- audio_type = os.path.basename(value).split(".")[-1].lower()
- if audio_type not in SUPPORT_AUDIO_TYPE_SETS:
- raise NotImplementedError(
- f'Not supported audio type: {audio_type}')
- if audio_type == "pcm":
- _type = "pcm"
-
+ audio_type = os.path.basename(value).lower()
+ if audio_type.rfind(".pcm") >= 0:
+ _type = "pcm"
func = DATA_TYPES[_type]
array = func(value)
if self.fs is not None and (name == "speech" or name == "ref_speech"):
@@ -336,11 +332,8 @@
# 2.a. Load data streamingly
for value, (path, name, _type) in zip(values, self.path_name_type_list):
if _type == "sound":
- audio_type = os.path.basename(value).split(".")[-1].lower()
- if audio_type not in SUPPORT_AUDIO_TYPE_SETS:
- raise NotImplementedError(
- f'Not supported audio type: {audio_type}')
- if audio_type == "pcm":
+ audio_type = os.path.basename(value).lower()
+ if audio_type.rfind(".pcm") >= 0:
_type = "pcm"
func = DATA_TYPES[_type]
# Load entry
@@ -392,3 +385,4 @@
if count == 0:
raise RuntimeError("No iteration")
+
--
Gitblit v1.9.1