Modify to 32K sample rate

Question

Modify to 32K sample rate

matt-audio opened this issue 3 years ago · comments

I tried to modified the DTLN model to 32K, but it always got error.
''----------------------------------------------------------------------------------------------------------------
File "/Library/Python/3.8/site-packages/tensorflow/python/keras/engine/training.py", line 1110, in fit
raise ValueError('Expect x to be a non-empty array or dataset.')
ValueError: Expect x to be a non-empty array or dataset.
''---------------------------------------------------------------------------------------------------------------

I changed the dataset to 32K, and modify self.fs = 32000, and all fs related stuffs.

Please help...

matt-audio · Answer 1 · Sat Jun 05 2021 15:42:14 GMT+0800 (China Standard Time)

I found the problem，Need to fix len_in_samples = 239872
but dont know why

Deleted user · Answer 2 · Mon Jul 26 2021 13:58:44 GMT+0800 (China Standard Time)

@matt-audio Hi，What is the noise reduction result of 32K model

matt-audio · Answer 3 · Mon Aug 02 2021 19:58:42 GMT+0800 (China Standard Time)

Just the same as the 16K model can be used ...

Deleted user · Answer 4 · Mon Aug 02 2021 21:41:08 GMT+0800 (China Standard Time)

您好，您的32k model 实时处理的音频有嗡嗡声嘛？🥲 大概就是人说话的时候不清晰！发自我的iPhone

…

在 2021年8月2日，19:58，matt-audio ***@***.***> 写道： Just the same as the 16K model can be used ... — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

matt-audio · Answer 5 · Tue Aug 03 2021 14:17:53 GMT+0800 (China Standard Time)

有些情况下有，感觉跟数据集有关系。。我目前也不知道是啥问题。。

________________________________ From: zql ***@***.***> Sent: Monday, August 2, 2021 9:41 PM To: breizhn/DTLN ***@***.***> Cc: matt-audio ***@***.***>; Mention ***@***.***> Subject: Re: [breizhn/DTLN] Modify to 32K sample rate (#42) 您好，您的32k model 实时处理的音频有嗡嗡声嘛？🥲 大概就是人说话的时候不清晰！发自我的iPhone

在 2021年8月2日，19:58，matt-audio ***@***.***> 写道： Just the same as the 16K model can be used ... — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#42 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ASHXU35E3YNIYXYALQAXIR3T22N77ANCNFSM46EEEDMA>.

Deleted user · Answer 6 · Tue Aug 03 2021 18:04:05 GMT+0800 (China Standard Time)

hi，能否告诉您的32k 模型参数？

…

在 2021年8月3日，14:18，matt-audio ***@***.***> 写道：有些情况下有，感觉跟数据集有关系。。我目前也不知道是啥问题。。 ________________________________ From: zql ***@***.***> Sent: Monday, August 2, 2021 9:41 PM To: breizhn/DTLN ***@***.***> Cc: matt-audio ***@***.***>; Mention ***@***.***> Subject: Re: [breizhn/DTLN] Modify to 32K sample rate (#42) 您好，您的32k model 实时处理的音频有嗡嗡声嘛？🥲 大概就是人说话的时候不清晰！发自我的iPhone > 在 2021年8月2日，19:58，matt-audio ***@***.***> 写道： > > Just the same as the 16K model can be used ... > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub, or unsubscribe. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#42 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ASHXU35E3YNIYXYALQAXIR3T22N77ANCNFSM46EEEDMA>. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.