Post your JAV subtitle files here - JAV Subtitle Repository (JSP)★NOT A SUB REQUEST THREAD★

[DScott]

[SPRD-678] Mother in Swimsuit Performed by Izumi Terasaki​

View attachment 3702288

I used WhisperJAV0.7 as the basis of this sub for SPRD-678 and I also used a couple of Sub downloads to fill in missing dialog several places. I also cleaned it up a bit and re-interpreted some of the meaningless/ "lewd-less" dialog. Again, I don't understand Japanese or Chinese so my re-interpretations might not be totally accurate but I try to match what is happening in the scene. Anyway, enjoy and let me know what you think.​

NIce work Chuckie. I have spent a lot of time experimenting with Whisper settings but I am much too lazy to try to guess what the dialog may be and then edit the .srt file. I run Whisper... Done! I hope that people appreciate your extra effort. The following are the word replacements that I have added to my clean_subs.py file which does post-processing and clean up. Maybe some of these may be of value, maybe not,,
(re.compile(r'(?i)\b(ramen|noodles|zamen|juice)\b'), 'cum'),
(re.compile(r'(?i)\bsoup\b'), 'sperm'),
(re.compile(r'(?i)\beggs\b'), 'semen'),
(re.compile(r'(?i)\bsleep\b'), 'blow a load of cum'),
(re.compile(r'(?i)\b(ferra|ferrachio|fera|fela?)\b'), 'sucking cock'),
(re.compile(r'(?i)\bblood\b'), 'cock juice'),
(re.compile(r'(?i)\b(chinko|chinpo)\b'), 'cock'),

 

[Chuckie100]

NIce work Chuckie. I have spent a lot of time experimenting with Whisper settings but I am much too lazy to try to guess what the dialog may be and then edit the .srt file. I run Whisper... Done! I hope that people appreciate your extra effort. The following are the word replacements that I have added to my clean_subs.py file which does post-processing and clean up. Maybe some of these may be of value, maybe not,,
(re.compile(r'(?i)\b(ramen|noodles|zamen|juice)\b'), 'cum'),
(re.compile(r'(?i)\bsoup\b'), 'sperm'),
(re.compile(r'(?i)\beggs\b'), 'semen'),
(re.compile(r'(?i)\bsleep\b'), 'blow a load of cum'),
(re.compile(r'(?i)\b(ferra|ferrachio|fera|fela?)\b'), 'sucking cock'),
(re.compile(r'(?i)\bblood\b'), 'cock juice'),
(re.compile(r'(?i)\b(chinko|chinpo)\b'), 'cock'),

Thanks...pretty close match to what I try to do manually. Except "Sleep" has always been a puzzler to me, will try your replacement in the future. Are you willing to share your "clean-up" .py file? I'm not familiar with running py files though, so may need "how to" explanations/directions to go along with it. Anyway, thanks.

 

[DScott]

Thanks...pretty close match to what I try to do manually. Except "Sleep" has always been a puzzler to me, will try your replacement in the future. Are you willing to share your "clean-up" .py file? I'm not familiar with running py files though, so may need "how to" explanations/directions to go along with it. Anyway, thanks.

Ok, some qualification is necessary. There are 3 sets of configurations here. My system is fairly powerful, I9-14900K (24Core/32Thread)/Nvidia 4080 Super (16G V-Ram)/96G RAM 6000mts/) Some of my settings willl have to be adjusted for a computer with less V-Ram in particular, like Float size or CPU threads but in essense the "slow" version yields the most accurate translation but is extremely slow. The setting that I mostly use is the "best fast version" which usually is pretty good at catching everything.

Most of the differences are within the transcribe.py file that determine how fast or slow or accurate or not. My slowest one does not employ VAD but yields the most accurate translation in my experience but I only use this if I find a subtitle that is glaringly missing text otherwise the fast one that employs VAD is usually pretty good. I don't think that there are any differences in the Whisper or Novad batch files and the clean_subs.py are the same in all three files. In essence as long is wisper.bat or NOVAD.bat are correctly pointing at your faster-whisper location you should only have to double click on either whisper.bat or novad.bat and the entire process should be automated. Now if you want to tinker with the settings you would do that in the transcribe.py file.

One other little thing that may confuse you is the japanese characters in my transcribe.py. Those are basically telling Whisper that it is translating JAV adult content.

THe files work like this: Whisper.bat initializes whisper, it calls to transcribe.py for processing information and then, after the .srt file is generated, then clean_subs kicks in to do cleanup/post processsing.. With regard to adding word replacements, just follow the syntax structure in my clean_subs.py file and you can add whatever changes you like.

Hopefully this all makes some kind of sense. Most of you know that i have a bad habit of rambling.

* re. "Sleep" being a puzzler to you. IT has always been a puzzler to me too and I just put that in there more for an inside joke because I couldn't quite figure what would be appropriate.

 

[Imscully]

Thanks...pretty close match to what I try to do manually. Except "Sleep" has always been a puzzler to me, will try your replacement in the future. Are you willing to share your "clean-up" .py file? I'm not familiar with running py files though, so may need "how to" explanations/directions to go along with it. Anyway, thanks.

Based on situations and context, I've come to believe that "Sleep" seems to be the equivalent of "horny".
Or, as Austin Powers might say....

 

[DScott]

Based on situations and context, I've come to believe that "Sleep" seems to be the equivalent of "horny".
Or, as Austin Powers might say....

View attachment 3702591

You had to do that didn you? YOU couldn't allow me to enjoy my day in relatively comatose state, oh no, you had to jarr all of my little neurons with this ,,, this... monstrosity.. Thanks a bunch! Oh, and thanks for the insight on "Sleep" I willl change this in my Clean_Sub file to reflect the newfound knowledge. I will either automate the replacement of sleep to be replaced with "Horny" or possibly "I Am Skullified",, I haven't made my mind up yet. Stay tuned/

 

[Imscully]

You had to do that didn you? YOU couldn't allow me to enjoy my day in relatively comatose state, oh no, you had to jarr all of my little neurons with this ,,, this... monstrosity.. Thanks a bunch! Oh, and thanks for the insight on "Sleep" I willl change this in my Clean_Sub file to reflect the newfound knowledge. I will either automate the replacement of sleep to be replaced with "Horny" or possibly "I Am Skullified",, I haven't made my mind up yet. Stay tuned/

Of course I did!!!
Here is another version.
It is farrrrrrr too reflective of my now long lost mojo-fueled self if I were to hit on a woman my own age.

 

[Chuckie100]

Based on situations and context, I've come to believe that "Sleep" seems to be the equivalent of "horny".
Or, as Austin Powers might say....

Yea, I think you are on to something...mainly, that "sleep" probably means different things depending on the context...not that that helps much though! Where the fuck is AI when you need him?! lol

 

[DScott]

Yea, I think you are on to something...mainly, that "sleep" probably means different things depending on the context...not that that helps much though! Where the fuck is AI when you need him?! lol

Here's a sign of the times. This morning I was making a comment on a video on Youtube. My phrasing didn't suit Youtube's Big Brother neurotic censor bot. So I went to Deepseek to give me an alternative phrasing that would razzle-dazzle Youtubes insane bot. It occurred to me that I was using Ai to outwit Ai on a subject that would not even cause a human to bat an eye. The world has truly gone loony-tunes.

 

[Prinsipe]

That version of the colab forces an old version of torch to fix an issue, but that stopped working because other packages are still up-to-date and that old version is no longer supported.

Going back to the original colab or using the pro version works better these days since they update everything, but some things tend to break from time to time as well. You can also use other more recent ones by mei2(or other people), they posts a link from time to time.

Removing the part that installs spleeter fixes things most of the time, since it's a separate package so compatibility with it is not checked with whisper and it's not used by default anyway(only is if you check source_separation).
So removing " spleeter" on the original or "!pip install spleeter" for the pro version in the "Setup Whisper" section after you press show code will usually do the trick.

Even the pro version is not functioning

I miss the Google Colab version of whisper before that has no error

It seems that 'Whisper with Silero VAD --maintenance release' and 'WhisperWithVAD PRO' have the same error. It seems that no one is maintaining the google colab of Whisper anymore. I hope someone can fix this and help us with the issue.

Thank you in advance for those who are willing to help.

 

[mei2]

Even the pro version is not functioning

I miss the Google Colab version of whisper before that has no error

It seems that 'Whisper with Silero VAD --maintenance release' and 'WhisperWithVAD PRO' have the same error. It seems that no one is maintaining the google colab of Whisper anymore. I hope someone can fix this and help us with the issue.

I just updated 'WhisperWithVAD PRO'. It should be working correctly now. Please test it when you can.

The root cause was a compatibility issue between Whisper, Spleeter and Python numpy that was triggered by an upgrade in Colab. Making all these to work together would be quite fragile and continues to break. For now, the best-fit solution I came up with is to deactivate Spleeter. I.e. the option for "Source Separation" is no longer available. If anyone still needs that option please let me know.

I have saved the older version as back up in the same folder if someone wants to work on it.

 

[Prinsipe]

I just updated 'WhisperWithVAD PRO'. It should be working correctly now. Please test it when you can.

The root cause was a compatibility issue between Whisper, Spleeter and Python numpy that was triggered by an upgrade in Colab. Making all these to work together would be quite fragile and continues to break. For now, the best-fit solution I came up with is to deactivate Spleeter. I.e. the option for "Source Separation" is no longer available. If anyone still needs that option please let me know.

I have saved the older version as back up in the same folder if someone wants to work on it.

What is the function of 'Spleeter' and 'Source Separation', will it affect much of the output quality?

Can I also request another favor? Can you also fix the 'Whisper with Silero VAD --maintenance release'? I have been using that colab for so long and I am very satisfied with its results.

Is WhisperWithVAD PRO better than Whisper with Silero VAD --maintenance release? What is the difference between the two? Sorry for too many questions, I am just really curious.

Thank you very much in advance for your help.

I really appreciate it.

 

[SamKook]

'Source Separation'(which relies on 'Spleeter') was off by default(but Spleeter had to be installed in case it got enabled) so unless you manually enabled it(which is unlikely), you've never used it before so quality will stay the same.

mei2 can say it better, but the difference between the 2 is the settings. You have more control with the pro and its defaults are better quality(but slower).

 

[kidbetin]

WhisperJAv Error

Message:
It looks like the pyloudnorm library is missing, which is causing an error during the import of the whisperjav module. I will add pyloudnorm to the installation command to fix this.

is it a colab error?

Thx in advance mei2

 

[mei2]

WhisperJAv Error

Message:
It looks like the pyloudnorm library is missing, which is causing an error during the import of the whisperjav module. I will add pyloudnorm to the installation command to fix this.

Thnaks for reporting it. I have fixed it now, it should run correctly.

 

[kidbetin]

HMNF-035

Just the Ayumi Shinoda clip.
The hard subs in the clip is not translated
Any subtitle addition is most welcome

563.4 MB file on MEGA

mega.nz

Posted subs as is, from mei2 whisperJAV colab

Thx mei2

 

[Novus.Toto]

START-346 Chase MINAMO! Sweat And Tears In This Big Chase SEX. 29 Facial Ejaculations In Total! Before You Know It, A Massive Bukkake Was Lifted! Special

• Code: START-346
• Release date: 2025-06-25
• Studio: SOD Create
• Label: START
• Tags: 3P, Bukkake, Facials, School Uniform, Solowork
• Actor: Eriguchi, Muscle Sawano
• Actress: MINAMO

1. Audio extracted and transcribed using WhisperWithVAD_Pro - both original speed audio and slowed audio.
2. The slowed down audio transcription was adjusted back to the correct timing.
3. Transcriptions cleaned to remove (most) hallucinations and other junk.
4. Both transcription versions were AI translated using DeepSeek v3 0324.
5. The "best" translation of each line from the two versions was used in the final translation.
6. Manually translated onscreen captions with Google Translate
7. Both VTT and SRT subtitle versions are in the zip. The translated onscreen captions aren't included in the SRT version.

There are several versions of this movie available for download - for some weird reason some are 30 fps and some are 29.97 fps. These subs should be synced to 29.97 fps. If you have a 30 fps version try re-synchronizing them to 30 fps. I suggest using Subtitle Edit to do that.

As always...
... I don't speak nor read Japanese so there are certain to be lots of mistakes.
It'd be extra-shiny if mistakes could be pointed out or corrected.

 

[Prinsipe]

I just updated 'WhisperWithVAD PRO'. It should be working correctly now. Please test it when you can.

The root cause was a compatibility issue between Whisper, Spleeter and Python numpy that was triggered by an upgrade in Colab. Making all these to work together would be quite fragile and continues to break. For now, the best-fit solution I came up with is to deactivate Spleeter. I.e. the option for "Source Separation" is no longer available. If anyone still needs that option please let me know.

I have saved the older version as back up in the same folder if someone wants to work on it.

There is still an error on WhisperWithVAD PRO, and I hope you can also fix the 'Whisper with Silero VAD --maintenance release'

I miss the 'Whisper with Silero VAD --maintenance release'

I hope someone can fix the errors in the two google colab

 

[SamKook]

That error is because you didn't execute the first block of code where the variables are defined. Just press play on the first block and it should work.

 

[Chuckie100]

I revisited my Sub for RCT-871, a classic Rocket mother-son vacation trip. I used WinX audio AI to try to bring out some of the voices at the beginning of the video and I also tried to clean it up a bit more., but multiple players are a bitch! Anyway enjoy. and let me know what you think.

 

[aronora]

May be there are earlier posted Eng Subs for older movie - UAAU-032 Hitomi Okumura - Son's Irresistible Morning Wood?
I have not seen it, and right now there is available quite good reduced mosaic version from that movie by portal NiceJav.
(Enhanced video quality > up to 1080p > NiceJav > /UAAU-032-UNCENSORED-LEAK )