search

qiuqiangkong / sampleRNN_acoustic_scene_generation

Other
13 stars 3 forks source link

Don't use condition on the sample level MLP ? #8

Open ghost opened 5 years ago

ghost commented 5 years ago

Hello, I was wondering why you didn't use the condition z (spkid in the code) with the input x in the sample level MLP. Is there a reason ? Intuitively, the condition (acoustic scene in your case) should have an impact on the sample level alignment as well as on the long term dependencies modeled by the RNN tier.

qiuqiangkong commented 5 years ago

Hi Rajesh,

It could be done condition on the sample level. But I did not do this at that time.

Best wishes,

Qiuqiang

From: RajeshMittal notifications@github.com Sent: 23 August 2019 15:04 To: qiuqiangkong/sampleRNN_acoustic_scene_generation sampleRNN_acoustic_scene_generation@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [qiuqiangkong/sampleRNN_acoustic_scene_generation] Don't use condition on the sample level MLP ? (#8)

Hello, I was wondering why you didn't use the condition z (spkid in the code) with the input x in the sample level MLP. Is there a reason ? Intuitively, the condition (acoustic scene in your case) should have an impact on the sample level alignment as well as on the long term dependencies modeled by the RNN tier.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fqiuqiangkong%2FsampleRNN_acoustic_scene_generation%2Fissues%2F8%3Femail_source%3Dnotifications%26email_token%3DADFXTSNFVU3T7B4IC3OEFWTQF7VAHA5CNFSM4IO776XKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HHBT6CQ&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C7ddbf1ef4195426813d308d727d2dea2%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637021658937422353&sdata=70%2BotArancxFuK97wmSpFAajGoHlyPQEsE3XSdubXVw%3D&reserved=0, or mute the threadhttps://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FADFXTSJT56ZKGULGKOOO5M3QF7VAHANCNFSM4IO776XA&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C7ddbf1ef4195426813d308d727d2dea2%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637021658937422353&sdata=JKe3wwolN%2BgIFr3bTlqyUJnYiEsdZ2T8ZSGniQub8hs%3D&reserved=0.

ghost commented 5 years ago

Hello, How would you do it ? The MLP part is most obscure to me. Would you do it before the convolution vector ? Thanks for replying so quickly

On Fri 23 Aug 2019 at 18:41, qiuqiangkong notifications@github.com wrote:

Hi Rajesh,

It could be done condition on the sample level. But I did not do this at that time.

Best wishes,

Qiuqiang

From: RajeshMittal notifications@github.com Sent: 23 August 2019 15:04 To: qiuqiangkong/sampleRNN_acoustic_scene_generation < sampleRNN_acoustic_scene_generation@noreply.github.com> Cc: Subscribed subscribed@noreply.github.com Subject: [qiuqiangkong/sampleRNN_acoustic_scene_generation] Don't use condition on the sample level MLP ? (#8)

Hello, I was wondering why you didn't use the condition z (spkid in the code) with the input x in the sample level MLP. Is there a reason ? Intuitively, the condition (acoustic scene in your case) should have an impact on the sample level alignment as well as on the long term dependencies modeled by the RNN tier.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub< https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fqiuqiangkong%2FsampleRNN_acoustic_scene_generation%2Fissues%2F8%3Femail_source%3Dnotifications%26email_token%3DADFXTSNFVU3T7B4IC3OEFWTQF7VAHA5CNFSM4IO776XKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HHBT6CQ&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C7ddbf1ef4195426813d308d727d2dea2%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637021658937422353&sdata=70%2BotArancxFuK97wmSpFAajGoHlyPQEsE3XSdubXVw%3D&reserved=0>, or mute the thread< https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FADFXTSJT56ZKGULGKOOO5M3QF7VAHANCNFSM4IO776XA&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C7ddbf1ef4195426813d308d727d2dea2%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637021658937422353&sdata=JKe3wwolN%2BgIFr3bTlqyUJnYiEsdZ2T8ZSGniQub8hs%3D&reserved=0

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/qiuqiangkong/sampleRNN_acoustic_scene_generation/issues/8?email_source=notifications&email_token=AKRVCYB56XTDPQ6ZX7KAYY3QGAHJ3A5CNFSM4IO776XKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5AXGMI#issuecomment-524383025, or mute the thread https://github.com/notifications/unsubscribe-auth/AKRVCYAEKEAJY2WPS4EQAKTQGAHJ3ANCNFSM4IO776XA .

qiuqiangkong commented 5 years ago

Hi, I have not tried and there is no plan to do this yet at moment.

Best wishes,

Qiuqiang

From: RajeshMittal notifications@github.com Sent: 24 August 2019 13:25 To: qiuqiangkong/sampleRNN_acoustic_scene_generation sampleRNN_acoustic_scene_generation@noreply.github.com Cc: Kong, Qiuqiang (PG/R - Elec Electronic Eng) q.kong@surrey.ac.uk; Comment comment@noreply.github.com Subject: Re: [qiuqiangkong/sampleRNN_acoustic_scene_generation] Don't use condition on the sample level MLP ? (#8)

Hello, How would you do it ? The MLP part is most obscure to me. Would you do it before the convolution vector ? Thanks for replying so quickly

On Fri 23 Aug 2019 at 18:41, qiuqiangkong notifications@github.com wrote:

Hi Rajesh,

It could be done condition on the sample level. But I did not do this at that time.

Best wishes,

Qiuqiang

From: RajeshMittal notifications@github.com Sent: 23 August 2019 15:04 To: qiuqiangkong/sampleRNN_acoustic_scene_generation < sampleRNN_acoustic_scene_generation@noreply.github.com> Cc: Subscribed subscribed@noreply.github.com Subject: [qiuqiangkong/sampleRNN_acoustic_scene_generation] Don't use condition on the sample level MLP ? (#8)

Hello, I was wondering why you didn't use the condition z (spkid in the code) with the input x in the sample level MLP. Is there a reason ? Intuitively, the condition (acoustic scene in your case) should have an impact on the sample level alignment as well as on the long term dependencies modeled by the RNN tier.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub< https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fqiuqiangkong%2FsampleRNN_acoustic_scene_generation%2Fissues%2F8%3Femail_source%3Dnotifications%26email_token%3DADFXTSNFVU3T7B4IC3OEFWTQF7VAHA5CNFSM4IO776XKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HHBT6CQ&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C7ddbf1ef4195426813d308d727d2dea2%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637021658937422353&sdata=70%2BotArancxFuK97wmSpFAajGoHlyPQEsE3XSdubXVw%3D&reserved=0>, or mute the thread< https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FADFXTSJT56ZKGULGKOOO5M3QF7VAHANCNFSM4IO776XA&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C7ddbf1ef4195426813d308d727d2dea2%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637021658937422353&sdata=JKe3wwolN%2BgIFr3bTlqyUJnYiEsdZ2T8ZSGniQub8hs%3D&reserved=0

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/qiuqiangkong/sampleRNN_acoustic_scene_generation/issues/8?email_source=notifications&email_token=AKRVCYB56XTDPQ6ZX7KAYY3QGAHJ3A5CNFSM4IO776XKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5AXGMI#issuecomment-524383025, or mute the thread https://github.com/notifications/unsubscribe-auth/AKRVCYAEKEAJY2WPS4EQAKTQGAHJ3ANCNFSM4IO776XA .

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fqiuqiangkong%2FsampleRNN_acoustic_scene_generation%2Fissues%2F8%3Femail_source%3Dnotifications%26email_token%3DADFXTSNGEHV5ZFKZPB7AM2LQGESDFA5CNFSM4IO776XKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5B7BVY%23issuecomment-524546263&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C899fc6620f494210317b08d7288e236e%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637022463253475901&sdata=6TP9JCadtbx19PSC%2BXMknDdYh00Ao2l3F3iilomh0Ys%3D&reserved=0, or mute the threadhttps://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FADFXTSNQPLZ4KJ2Z3ZMJG53QGESDFANCNFSM4IO776XA&data=02%7C01%7Cq.kong%40surrey.ac.uk%7C899fc6620f494210317b08d7288e236e%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637022463253485898&sdata=LBleQRAVXtsY2w%2BAEyAKdivWYpUca%2FGCu9YItVlZErA%3D&reserved=0.