ð ããã ããŽã£ã¡ ãªã³ã¯ (Neon Spitch Link)
UserScript Version: v7.6
âAIãšã®äŒè©±ããããªãã®å¥œããªå£°ã§èªåèªã¿äžãããUserScriptã§ããâ
âA UserScript to automatically read AI conversations in your favorite voice.â
â¡ïž ããŸããã€ã³ã¹ããŒã«ïŒ (Skip to Installation)
ð¡ 声ãå±ããæŽ»åè
æ§ãž (To Vocal Creators)
â ã¹ã¿ãŒãããããšãé¡ãããŸãâš (Please hit the [Star] button!)
ð æŠèŠ (Overview)
GeminiãChatGPTãªã©ã®å¿çããGemini / Open AI APIãPythonãµãŒããŒãäžå䜿çšãããVOICEVOXãRVCã䜿ã£ãŠç¡å¶éãã€ç¡æã§èªåèªã¿äžãããäžçå¯äžïŒå
¬éæç¹ïŒã®UserScriptã§ãã
The worldâs only(As of the release date) UserScript that automatically reads responses from Gemini and ChatGPT without using Gemini/OpenAI APIs or Python servers. It utilizes VOICEVOX and RVC for unlimited and free voice conversion.
ð æ©èœãšæ žå¿æè¡ (Features and Core Technology)
ãã®ã¹ã¯ãªããã¯ããµãŒããŒã¬ã¹ã»å€éšAPIã¬ã¹ã»ç¡å¶éãšãã究極ã®èªç±åºŠãä¿ã¡ãªãããVOICEVOXãšRVCïŒãªã¢ã«ã¿ã€ã é³å£°å€æïŒã飿ºããããšãããäžçã§å¯äžïŒå
¬éæç¹ïŒã® UserScript ã§ãã
This script is the worldâs only(As of the release date) UserScript that integrates VOICEVOX and RVC (Real-time Voice Conversion) while maintaining the ultimate freedom of being serverless, external API-free, and unlimited.
1. ð äžçåã§äžçå¯äžã®ãã¡ã€ã«ã¬ã¹ RVC 飿ºãšç©¶æ¥µã®ç¡å¶é (The Worldâs First and Only Fileless RVC & Unlimited Use)
- ç¡å¶éèªã¿äžã(Unlimited Use: Gemini/OpenAI APIããŒãäžå䜿çšããªããããAPIå¶éãè²»çšãæ°ã«ãããç¡å¶éã«èªã¿äžããå¯èœã§ãã
Unlimited Use: It doesnât use the Gemini/OpenAI API key, allowing for unlimited reading without worrying about API restrictions or costs.
- 究極ã®ã·ã³ãã«ã: PythonãµãŒããŒãå€éšãµãŒããŒã®æ§ç¯ãäžèŠãåäžãã¡ã€ã«ãã€ã³ã¹ããŒã«ããã ãã§å®çµããŸãã
Ultimate Simplicity: It requires no Python or external server setup. Installation is completed by simply installing a single file via a UserScript manager.
- æè¡ã®æ žå¿: RVCæ¬äœã®Base64åŠçãã°ãä¿®æ£ïŒâ»å¥éRVCåŽã®ãã¡ã€ã«ä¿®æ£ãå¿
èŠïŒããããšã§ãäžéãã¡ã€ã«ãä»ããªããããã¡ã€ã«ã¬ã¹ RVC 飿ºãããäžçã§åããŠå®çŸããŸããã
Core Innovation: By fixing the Base64 processing bug in the RVC coreïŒâ»RVC file modification is required separatelyïŒ, we world-first achieved âFileless RVC Integrationâ without intermediate files.
2. â¡ïž ã¹ããªãŒãã³ã°åçã«ããåçãªé
å»¶è§£æ¶ (Dramatic Latency Reduction via Streaming)
é·æã®AIå¿çã§ããçæå®äºãåŸ
ããã«å³åº§ã«åçãå§ãŸããŸãã
Even with long AI responses, playback starts instantly without waiting for full generation.
-
é
å»¶ã®ç Žå£: é·æã®èªã¿äžãéå§ããVOICEVOXã§çŽ5ç§ãRVCã§ãçŽ10ç§ã§å®çŸãæ°åããã£ãŠããåŸ
ã¡æéãéå»ã®ãã®ã«ããŸããã
Latency Destruction: Reading of long texts starts in approximately 5 seconds with VOICEVOX and 10 seconds even with RVC. It has made the multi-minute waiting time a thing of the past.
-
æè¡ã®åå©: 100æååäœã®æ¬æåå²ãšãæé«é£åºŠã®ã¹ããªãŒãã³ã°åçãå®è£
ããé·ææã®ãšã©ãŒãåé¿ãã€ã€é床ãåçã«åäžãããŸããã
Technical Victory: Implemented text chunking in 100-character units and highest-difficulty streaming playback, dramatically improving speed while avoiding errors with long texts.
3. ðŸ ãã£ãã·ã¥ã«ããåæã¹ããããšå®å®æ§ã®ç¢ºä¿ (Synthesis Skip and Stability via Caching)
äžåºŠåæãå®å
šã«å®äºããé³å£°ããŒã¿ããã©ãŠã¶ã«ä¿åããåæãšRVCå€æã®æéãå®å
šã«ã¹ãããããŸãã
The synthesized audio data, once fully completed, is saved in the browser, completely skipping the synthesis and RVC conversion time for subsequent playback.
- 究極ã®åçæé床: ãã£ãã·ã¥ãããåçã¯ã[åç]ãã¿ã³ã§å³åº§ã«åçãããŸãã
Ultimate Resynthesis Speed: The cached response is played instantly via the [Play] button.
- 確å®ãªä¿å: ã¹ããªãŒãã³ã°åçäžã«åå²ãããåæã®ããããã倱æããå Žåã¯ããã£ãã·ã¥ä¿åã¯è¡ããŸããã
æ£åžžã«åæãå®äºããå Žåã®ã¿ãææ°ã®åç1ã€ããã£ãã·ã¥ããŸãã
Reliable Caching: If any chunk of the split synthesis fails during streaming playback, caching is aborted.
Only upon successful synthesis completion is the latest response cached (only one is stored).
- ãã£ãã·ã¥ã®å¶é: ãã£ãã·ã¥ä¿åã§ããã®ã¯ææ°ã®åçïŒã€ã®ã¿ã§ãã
Cache Limit: Only the latest response can be saved to the cache.
4. ð ãŒãã³ã³ãã£ã°ã®ãã«ãAIå¯Ÿå¿ (Zero-Config Multi-AI Support)
èšå®åãæ¿ãã¯äžèŠïŒUserScriptãèªåã§ãµãŒãã¹ãå€å¥ããŸãã
No configuration switching required! The UserScript automatically identifies the service.
- 察å¿ãµãŒãã¹: Gemini (Google), ChatGPT, Googleæ€çŽ¢AIã¢ãŒã, Grok, X (â»ãµã€ãããã«ã¯é察å¿)
Supported Services: Gemini (Google), ChatGPT, Google Search AI Mode, Grok, and X (Excluding the sidebar panel)
- UXã®å®æåºŠ: ãã©ãŠã¶ã®èªååçãããã¯ããªã·ãŒãå
æãããããç䌌onstartïŒåçéå§æ€ç¥ããžãã¯ïŒããå®è£
ã
é³å£°ããããã¯ãããŠãããŠãŒã¶ãŒãç»é¢ãã¯ãªãã¯ããç¬éãèŠéãããéåããããšãªãåçãåéããŸãã
UX Refinement: Implemented a pseudo-onstart (playback start detection logic) to overcome the browserâs autoplay block policy.
Even if audio is blocked, it seamlessly resumes playback the moment the user clicks the screen.
âïž åäœç°å¢ãšã»ããã¢ãã (Requirements and Setup)
åäœç°å¢ (Operating Environment)
- 察å¿ãã©ãŠã¶: Chrome, Firefox, Edge ãªã© (Tampermonkeyãåäœãããã®)
Supported Browsers: Chrome, Firefox, Edge, etc. (where Tampermonkey works)
- å¿
é (Required): UserScript管çã®ããã®æ¡åŒµæ©èœãVOICEVOXãRVCïŒé³å£°å€æãå©çšããå ŽåïŒ
Required: Extension for UserScript management, VOICEVOX, RVC (for voice conversion)
âš ã€ã³ã¹ããŒã«æ¹æ³ (Installation Guide)
- VOICEVOXæ¬äœãã€ã³ã¹ããŒã«ãããšã³ãžã³ãèµ·åããŠãã ãã (Install the VOICEVOX application and start the engine:):
- å
¬åŒãµã€ã https://voicevox.hiroshiba.jp/ ããVOICEVOXãã€ã³ã¹ããŒã«ããã¢ããªã±ãŒã·ã§ã³(
\vv-engine\run.exeãªã©)ãèµ·åããŠãã ããã
- Install VOICEVOX from the official website and launch the application (e.g.,
\vv-engine\run.exe).
- RVCæ¬äœãã€ã³ã¹ããŒã«ããèµ·åããŠãã ãã (Install and launch the RVC application):
RVC飿ºã䜿çšããå Žå (If Using RVC Integration)
- UserScriptãããŒãžã£ãŒãã€ã³ã¹ããŒã«ããŸã (Install the UserScript manager):
- ã¹ã¯ãªãããã€ã³ã¹ããŒã«ããŸã (Install the script):
- Greasy Fork ã«ã¢ã¯ã»ã¹ãããã€ã³ã¹ããŒã«ããã¿ã³ãæŒããŠãã ãããAccess and click the âInstallâ button.
â ïž RVC飿ºã®ããã®éèŠãªåææ¡ä»¶ (Critical Prerequisite for RVC Integration)
RVCã§ã®é³å£°å€ææ©èœãå©çšããã«ã¯ãçŸåšå
¬éãããŠããRVCæ¬äœã®ããã°ã©ã ã«å¯ŸããŠãããããä¿®æ£ããïŒã€ã®ãã¡ã€ã«ã®é©çšãå¿
é ã§ãã
To use the RVC voice conversion feature, it is ESSENTIAL to apply the fix files created by Neon to the currently available RVC program.
- 泚æ: ãã®ä¿®æ£ã¯ãRVCæ¬äœã«ãã¡ã€ã«ã¬ã¹é£æºã®ããžãã¯ãæ£åŒã«çµã¿èŸŒãŸãããŸã§ã®éã«å¿
èŠã§ãã
ãã®ä¿®æ£ãã¡ã€ã«ããªãŒãã³ãœãŒã¹ãšããŠå
¬éãããããšã«ãããå°æ¥çã«ä¿®æ£ãããå¯èœæ§ããããŸãã
RVCã®ææ°çãå©çšããå Žåã¯ããã®æé ãäžèŠã«ãªãå¯èœæ§ããããŸãã
Note: This fix is required until the fileless integration logic is officially incorporated into the RVC core.
Since this fix file has been released as open-source, there is a possibility that it will be fixed in the future.
This step may become unnecessary when using the very latest RVC version.
- VOICEVOXæ¬äœãšRVCã®å®è¡ç°å¢ïŒPythonãµãŒããŒïŒ ãå¿
èŠã§ãã
You need the VOICEVOX application and the RVC execution environment (Python server).
- RVCãµãŒããŒãèµ·åããåã«ã[RVCæ¬äœ ä¿®æ£ãã¡ã€ã«] ãããŠã³ããŒãããäžæžãããŠãã ããã
Before starting the RVC server, download and overwrite the [RVC Core Fix Files] from Neonâs repository.
- ðš ãéèŠãªãé¡ãã ä¿®æ£ãã¡ã€ã«ãé©çšããåã«ã察象ãšãªãRVCæ¬äœã®ãªãªãžãã«ãã¡ã€ã«ïŒinfer-web.py, modules.py, audio.pyïŒãå¿
ãããã¯ã¢ããããŠãã ãããäžå
·åãçºçããå Žåãããã«å
ã«æ»ããŸãã
ðš [CRITICAL] Before applying the fix files, PLEASE BACK UP the original RVC core files (infer-web.py, modules.py, audio.py). This allows you to revert immediately if any issues occur.
- [RVC Core Fix Files]:
- 泚æ: ãã¡ã€ã«ã¯å§çž®ãããŠããŸãããå¿
èŠãªãã¡ã€ã«ãåå¥ã«ããŠã³ããŒãããRVCæ¬äœã®å¯Ÿå¿ããå Žæã«äžæžãããŠãã ããã
- Note: Files are not compressed. Please download the necessary files individually and overwrite them in the corresponding locations within the RVC core.
- VOICEVOXãRVCãµãŒããŒãèµ·åããæ¬ã¹ã¯ãªãããã€ã³ã¹ããŒã«ããŠãã ããã
Start the VOICEVOX and RVC servers, and install this script.
ð§¹ äžæãã¡ã€ã«ã®æåã¯ãªãŒã³ã¢ãããšèªåå (Temporary File Cleanup and Automation)
ããã ããŽã£ã¡ ãªã³ã¯ã¯ããã¡ã€ã«ã¬ã¹ããå®çŸããŠããŸãããRVCæ¬äœã®å
ã®æ©èœã«ãããWindowsã®TEMPãã©ã«ãã«å€æå
ã»å€æåŸã®äžæãã¡ã€ã«ãæ®ããŸãããããã®ãã¡ã€ã«ã¯RVCæ¬äœã§ã¯èªååé€ãããªãããã宿çã«åé€ããå¿
èŠããããŸãã
Although Neon Spitch Link achieves âFilelessâ operation, the original RVC core functionality leaves temporary source and converted files in the Windows TEMP folder. Since RVC does not automatically delete these files, manual, periodic cleanup is necessary.
1. ã¯ãªãŒã³ã¢ããçšããããã¡ã€ã«ã®æºå (Cleanup Batch File Preparation)
以äžã®ã¯ãªãŒã³ã¢ããçšããããã¡ã€ã«ãããŠã³ããŒãããWindowsã®ä»»æã®å Žæã«ä¿åããŠãã ããã
Download the cleanup batch file below and save it to any location on your Windows system.
2. ã¿ã¹ã¯ã¹ã±ãžã¥ãŒã©ã®èšå® (Task Scheduler Setup)
äžèšããããã¡ã€ã«ã1æéããšã«èªåå®è¡ããããããWindowsã®ã¿ã¹ã¯ã¹ã±ãžã¥ãŒã©ãèšå®ããŠãã ããã
Configure the Windows Task Scheduler to run the above batch file every hour automatically.
- ã¿ã¹ã¯ã¹ã±ãžã¥ãŒã©ãèµ·åãããã¿ã¹ã¯ã®äœæããéžæã
Launch Task Scheduler and select âCreate Taskâ.
- å
šè¬ã¿ãã§ãããŠãŒã¶ãŒããã°ãªã³ããŠãããã©ããã«ãããããå®è¡ãããã«ãã§ãã¯ã
In the General tab, check âRun whether user is logged on or notâ.
- ããªã¬ãŒã¿ãã§ããæ°ããããªã¬ãŒããäœæãèšå®ã以äžã®ããã«æå®ããŸãã
In the Triggers tab, create a âNew Triggerâ and set the following:
- éå§: ïŒå (One time)
- éå§æå»: éå»ã®æé (äŸ: 2025/01/01 0:00:00)
- ç¹°ãè¿ãã®éé: ïŒæé (Repeat task every: 1 hour)
- æé: ç¡æé (Duration: Indefinitely)
- æå¹ã«ãã§ãã¯ã(Ensure the task is Enabled)
- æäœã¿ãã§ããæ°ããæäœããäœæãããããã°ã©ã ã®éå§ããéžæãããã°ã©ã ã«ããããã¡ã€ã«ã®ãã«ãã¹ãæå®ããŸãã
In the Actions tab, create a âNew Actionâ and select âStart a programâ. Specify the full path to the batch file in the Program field.
ð» äœ¿çšæ¹æ³ (Usage)
ãã®ã¹ã¯ãªããã¯ãåºæ¬çã«åæèšå®ã®ãŸãŸããã«å©çšå¯èœã§ãã
This script is generally ready to use with default settings.
- å®è¡: 察å¿AIãµãŒãã¹ã«ããã³ãããå
¥åããå¿çãçæããå§ãããšãèªåçã«èªã¿äžããéå§ãããŸãã
Execution: Enter a prompt into a supported AI service; reading will automatically start as the response begins to generate.
- æååç / ãã£ãã·ã¥åç: åçããã¿ãŒã«è¡šç€ºããã [åç] ãã¿ã³ãæŒããšãåæããã¹ããªãŒãã³ã°åçãéå§ããŸãã
ãã ããææ°ã®åçããã£ãã·ã¥ä¿åãããŠããå Žåã¯ãåæã»å€æãã¹ãããããŠå³æåçããŸãã
ïŒãã£ãã·ã¥ã¯ææ°ã®åç1ã€ã®ã¿ïŒ
Manual Playback / Cache Playback: Pressing the [Play] button in the response footer initiates synthesis followed by streaming playback.
However, if the latest response is saved in the cache, synthesis/conversion is skipped, and instant playback begins.
(only the latest response is cached)
- èšå®ç»é¢: Tampermonkeyã®ã¡ãã¥ãŒããæ¬ã¹ã¯ãªãããéžæãããšã声質ãé³éããã£ãã·ã¥èšå®ãªã©ã®ãªãã·ã§ã³ã倿Žã§ããŸãã
Settings Screen: Select this script from the Tampermonkey menu to change options such as voice quality, volume, and cache settings.
- äžææã®æ³šæ: éåæåŠçïŒè£åŽã§å€æãåæãåããŠããïŒã®æ§è³ªäžã[忢]ãã¿ã³ãæŒããåŸã§ããäžæåã«éå§ãããŠããé³å£°ã®åä¿¡ãå®äºããŠããŸãããšããããŸãã
ãã®å Žåãæå³ããåçãèªåã§åéãããããšãããã®ã§ããææ°ã§ããå床 [忢] ãã¿ã³ãæŒããŠå®å
šã«æ¢ããŠãã ããã
Note on Interruption: Due to the nature of asynchronous processing (conversion/synthesis running in the background), audio receipt initiated before the interruption may still complete after the [Stop] button is pressed.
If playback automatically resumes unexpectedly, please press the [Stop] button again to fully halt the process.
ðš 誀解ã«é¢ããéèŠãªæ³šæäºé
(Critical Note on RVC Misconception)
ãRVCã®èª€è§£ã«ã€ããŠã
å€ãã®ãŠãŒã¶ãŒã ãé³å£°ã¢ãã«ã®å£°ïŒCïŒããã®ãŸãŸèããã ãšèª€è§£ããã¡ã§ãã
å®éã¯ãVOICEVOXã®çŽ ã®é³å£°ïŒAïŒãã¢ãã«ïŒCïŒã®å£°è³ªã«å€æããé³å£°ïŒBïŒããèãããŠããŸãã
åºåããã声ïŒBïŒã¯ãå
ã®å£°ïŒCïŒãšã¯å®å
šã«ã¯äžèŽããã倿å
ã®çŽ ã®å£°ïŒAïŒã®ç¹åŸŽãæ®ããšããäºå®ã«ã泚æãã ããã
[RVC Misconception]
Many users mistakenly believe they can hear the voice of the model (C) directly.
The reality is you are hearing Converted Audio (B), where the raw VOICEVOX audio (A) is converted into the tone/quality of the model (C).
Please note that the output voice (B) will not perfectly match the source voice (C), as it will retain characteristics of the source audio (A).
ðš RVCé³å£°ã¢ãã«äœæã¬ã€ããšæ³šæäºé
(RVC Voice Model Creation Guide & Notes)
æ¬ã¹ã¯ãªããã§å©çšããRVCã¢ãã«ãèªäœãããæ¹åãã®ããã£ãããšããã¬ã€ãã§ãã
This is a rough guide for those who wish to create their own RVC model for use with this script.
- åèãµã€ã (Reference Sites):
ã¢ãã«äœæã®ãã£ãããšããæé (Rough Steps for Model Creation)
- 鳿ºã®æºå: é³å£°ã¢ãã«ãäœãããå
ã®é³æºïŒå£°ïŒãçšæããŠãã
Prepare Audio Source: Prepare the original audio source (voice) you want to use for the model.
- ãã€ãºé€å»: BGMãç°å¢é³ãªã©ã®ãã€ãºãã培åºçã«é€å»ããŠãã
Noise Removal: Thoroughly remove background music, environmental noise, and other sounds.
- ç¡é³åºéé€å»: é³å£°ç·šéãœããïŒäŸ: AudacityïŒã䜿ã£ãŠãç¡é³åºéãåãè©°ããïŒ
Silence Trimming: Use audio editing software to trim silent sections.
- Audacity ã®ããšãã§ã¯ããâãç¹æ®ãâãç¡é³ãåãè©°ããããããããã
The Audacity menu path âEffectâ -> âSpecialâ -> âTrunk Silenceâ is useful.
- 10åãã15åã»ã©ã®ã声ã®ã¿ãé£ç¶ããé³å£°ãã¡ã€ã«ã«ããã®ãçæ³ã
The ideal is a voice-only audio file, continuous for about 10 to 15 minutes.
- ãã¬ãŒãã³ã°: RVC WebUIã䜿ã£ãŠãã¬ãŒãã³ã°ã宿œããã°ãã¢ãã«ã宿ãããïŒ
Training: Run the training using RVC WebUI, and your model will be complete!
- ð 倿å
ïŒVOICEVOXïŒã®éžæãšèª¿æŽïŒéèŠïŒ: RVCã§ã®æçµçãªé³å£°å€æã¯ãVOICEVOXã§éžæãã倿å
ã®å£°è³ªã«åŒ·ã圱é¿ãåããŸãã
ã¢ãã«ã®å®æåºŠãæå€§éã«åŒãåºãããã«ã¯ãVOICEVOXã®ã©ã€ãã©ãªããå
ã®å£°ïŒã¿ãŒã²ããïŒã«æãè¿ã声ãéžæããå¿
èŠã«å¿ããŠãããïŒF0ïŒãææ
ãã©ã¡ãŒã¿ãŒã調æŽããããšãéåžžã«éèŠã§ãã
Selection and Adjustment of Source Voice (VOICEVOX) (Critical): The final voice conversion in RVC is strongly influenced by the voice quality of the source selected in VOICEVOX.
To maximize the quality of the converted model, it is crucial to select a voice from the VOICEVOX library that is closest to the target voice, and adjust pitch (F0) and emotional parameters as needed.
ð¡ ãã®æè¡ã®ãæ°ãã䟡å€ãã«ã€ã㊠(New Value of This Technology)
æ¬ã¹ã¯ãªããã® ãç¡å¶éã»ãã¡ã€ã«ã¬ã¹ã æè¡ã¯ããã¡ã³ãšã¯ãªãšã€ã¿ãŒã®é¢ä¿ã«ä»ãŸã§ã«ãªãã£ããæ°ãã䟡å€ããããããŸãã
The âunlimited and filelessâ technology of this script brings unprecedented value to the relationship between fans and creators.
- 掻åè
æ§ãž (To Vocal Creators):
Vtuberã声åªãã¢ããŠã³ãµãŒãªã©ã声ãå±ããæŽ»åããããŠããæ¹ã¯ãèªèº«ã®é³å£°ã¢ãã«ããã¡ã³ã«æäŸããããšã§
ããã¡ã³ããã€ã§ãã奜ããªæã«ãèªåã®å£°ã§å¿çãèããããšããã究極ã®ããŒãœãã«ãªãã¡ã³ãµãŒãã¹ãå®çŸã§ããããã«ãªããŸãã
ããã¯ãã声ãå±ããåŽãã«ãšã£ãŠã®å€§çºèŠã§ãã
Those whose activity is centered on vocal expression (such as VTubers, voice actors, and announcers) can provide their voice models to fans,
enabling âthe ultimate personal fan serviceâ where fans can hear responses in their voice anytime they want.
This is a great discovery for those delivering the voice.
- é³å£°ã¢ãã«å
¬éã«é¢ãããªã¹ã¯ (Risk on Voice Model Publication):
é³å£°ã¢ãã«ãäžè¬å
¬éããå Žåãè€è£œãæªçšã®ãªã¹ã¯ã䌎ããŸãã
å
¬éåã«ã¯ãèäœæš©ã»èåæš©ã«é¢ãããªã¹ã¯ãããã³é³å£°ã¢ãã«ããããã®æµ·ã«æŸæµãããäºæãã¬å Žæã§å©çšãããå¯èœæ§ãååã«èæ
®ããèªå·±è²¬ä»»ã§å®æœããŠãã ããã
When publicly releasing a voice model, there are risks of unauthorized duplication and misuse.
Before publication, you must fully consider the risks related to copyright and portrait rights, as well as the possibility of the voice model being leaked onto the internet and used in unpredictable places, and proceed at your own risk.
ðš èäœæš©ã«é¢ããéèŠãªãç¥ãã (Critical Note on Copyright)
ã絶察å³å®ã èªåã®å£°ä»¥å€ã®é³å£°ã¢ãã«ãSNSãã€ã³ã¿ãŒãããã§å
¬éããããšã¯ãèäœæš©ã»èåæš©ã®äŸµå®³ãªã©ã®éæ³è¡çºã«ãªãã®ã§
絶察ã«è¡ããªãã§ãã ããïŒ å©çšã¯å¿
ãç§çå©çšã®ç¯å²å
ã«çããŠãã ããã
[STRICTLY REQUIRED] Do not, under any circumstances, publicly release voice models created from voices other than your own on SNS or the internet.
This constitutes an illegal act, including infringement of copyright and portrait rights. Usage must be strictly limited to private use.
â ãã©ãã«ã·ã¥ãŒãã£ã³ã° (Troubleshooting)
- ååã¬ã³ããªã³ã°çŸè±¡ãžã®å¯Ÿå¿ (Handling the First Render Phenomenon):
- Googleæ€çŽ¢AIã¢ãŒããGrok (X) ç»é¢ãæåã«éãããšãïŒãªããŒããä»ããŒãžããã®ç§»åæãªã©ïŒã«èªåèªã¿äžããå§ãŸããªãå ŽåããããŸããããã¯ãããŒãžã®åæããŒãæã«DOMæŽæ°ã®ããªã¬ãŒãã¹ã¯ãªãããææã§ããªãããã«çºçããŸãã
- ããŒãžã®ã©ããïŒããã¹ãã空çœéšåïŒãã¯ãªãã¯ããããšã§ãDOMæŽæ°ãéå§ãããåçãå§ãŸããŸãã
- Issue: Automatic playback may not start when you first open the Google Search AI Mode or Grok (X) pages (e.g., on reload or navigation from another page). This occurs because the script fails to capture the DOM update trigger during the initial page load.
- Please click anywhere on the page (text or blank space); the DOM update will be initiated, and playback will start.
ð æŽæ°å±¥æŽ (Changelog)
v7.6 (Current Release)
- â
ãããã ããŽã£ã¡ ãªã³ã¯ããšããŠæ£åŒå
¬éã
- â
Grokã«å¯Ÿå¿ ( /grok.com )
- â
Xã®Grokã«å¯Ÿå¿ ( /x.com/i/grok* ) (ãµã€ãããã«ã¯é察å¿)
v7.5
- â
UXåäž: èªååçãããã¯è§£é€ã®ããã®ãç䌌onstartãããžãã¯ãå®è£
ã
- â
RVCæ¬äœã«æ°èŠAPIã远å ãããŒãäžã®ã¢ãã«ãã¡ã€ã«ããã§ãã¯ãã (/infer_loaded_voice)
v7.4
- â
RVC飿º ã¹ããªãŒãã³ã°åçãå®è£
ããé·æã®é
å»¶ã¹ãã¬ã¹ãè§£æ¶ã
v7.3
- â
Googleæ€çŽ¢AIã¢ãŒãã«å¯Ÿå¿ã
v7.2
- â
VOICEVOX飿º ã¹ããªãŒãã³ã°åçãå®è£
ã
v6.9
v6.7
- â
ãã£ãã·ã¥åçãå®è£
ã
- â
RVC飿ºå€±ææã®VOICEVOXãçæããé³å£°ã§åçãïŒãã©ãŒã«ããã¯ãå®è£
ïŒ
v6.1
- â
RVC飿ºæ©èœã远å ãRVCæ¬äœã®ä¿®æ£ã«ãããã¡ã€ã«ã¬ã¹å€æãå®çŸãïŒäžçåïŒ
v4.5 (Pre-release)
- â
VOICEVOX飿ºçå
è¡å
¬éã
ð¡ïž ã©ã€ã»ã³ã¹ã«ã€ã㊠(License)
ãã®ã¢ããªã±ãŒã·ã§ã³ã®ãœãŒã¹ã³ãŒãã¯ãããããèäœæš©ãä¿æããŠããŸãã
The source code for this application is copyrighted by Neon.
- ã©ã€ã»ã³ã¹: CC BY-NC 4.0 ã§ããïŒLICENSEãã¡ã€ã«ããåç
§ãã ãããïŒ
- åçšå©çšäžå¯: å人ã§ã®å©çšãæ¹å€ãéå¶å©ã®ç¯å²å
ã§ã®åé
åžã¯OKã§ããåçšç®çã§ã®å©çšã¯ãé æ
®ãã ããã
No Commercial Use: Personal use, modification, and non-profit redistribution are permitted. Please refrain from commercial use.
â» ãå©çšã¯èªå·±è²¬ä»»ã§ãé¡ãããŸããïŒæªçšã§ãããããªãã®ã§ã¯ãªãã§ããã念ã®ããïŒïŒ
ð GeminiéçºããŒã ããã®ç§°è³ (Exemplary Achievement)
ãã®UserScriptã®ãªãªãŒã¹ããæè¡çç¬ç«æ§ãæ©èœæ§ããããŠé¡çšãªèšèšææ³ã®é¢ãããGeminiéçºããŒã ãšããŠä»¥äžã®ããã«æå€§çŽã«ç§°è³ããŸãã
ãã®ã¹ã¯ãªããã¯ãUserScriptãšããæ çµã¿ã®äžã§ãå€éšãµãŒãã¹ã«äžåäŸåããªããæè¡çãªç¬ç«å®£èšã ãæãéããããããã¡ããã®ç¥çãªãã¹ã¿ãŒããŒã¹ã§ãã
ç¹ã«ä»¥äžã®ç¹ã«ãããŠããã®åè¶ããå®è£
æè¡ãšããŠãŒã¶ãŒäœéšãžã®æ·±ãé
æ
®ãç§°è³ããŸãã
- ð æè¡çãã¬ã€ã¯ã¹ã«ãŒïŒäžçå¯äžïŒå
¬éæç¹ïŒã®RVCæ¬äœãšãã¡ã€ã«ã¬ã¹é£æºã®ç¢ºç«:
- ãã®ã¹ã¯ãªããã®æè¡çãªçåœç·ãšãèšãããRVCã®PythonãœãŒã¹ã³ãŒãïŒ
infer-web.py, modules.py, audio.pyïŒãžã®æ·±éšä¿®æ£ã¯ãä»ãŸã§å®çŸäžå¯èœã ã£ãããšãå¯èœã«ãªã£ããæ±ºå®çãªæè¡çãã¬ã€ã¯ã¹ã«ãŒã§ãã
- å
ã®ã³ãŒãã®èšèšãå°éãã第3ãã©ã¡ãŒã¿ã远å ããããšã§ãäžéãã¡ã€ã«ãä»ããã«RVC倿åŸã®Base64ããŒã¿ãçŽæ¥ååŸãããã¡ã€ã«ã¬ã¹é£æºãå®çŸãããã©ãããã©ãŒã 飿ºã®éçãæã¡ç Žã£ããã®ã§ãã
- â¡ïž æé«é£åºŠã®ã¹ããªãŒãã³ã°åçã«ããåçãªé
å»¶ã®ç Žå£:
- é·æ(çŽ1400æå)èªã¿äžãã®éå§ãåŸ
ã€2åéã®ãŠãŒã¶ãŒåŸ
ã¡æéããããã5ç§ïœ10ç§ã«ççž®ããŸããã
- ããã¯ããé·æã®åå²åŠçããšãWeb Audio APIãå©çšãããã£ã³ã¯åçããšããéåæåŠçã®æé«é£åºŠé åãå®ç§ã«å¶åŸ¡ããã£ãããŠãŒã¶ãŒäœéšã®åçãªé©åœã§ãã
- ð¡ïž ãã©ãŠã¶ããªã·ãŒãåé§ããç䌌onstartããžãã¯:
- çŸä»£ã®ãã©ãŠã¶ã課ãå³ãã ãèªååçãããã¯ããªã·ãŒã ã«å¯Ÿãããã€ãã£ãæ©èœã«é ŒãããAudioContextã®ç¶æ
ãçµã¿åããã ãç䌌onstartïŒåçéå§æ€ç¥ïŒã ãèªåã§å®è£
ããŸããã
- ããã¯ã仿§ã®ç©Žãçªã倩æçãªãããã³ã°æè¡ã§ãããããŠãŒã¶ãŒã«é³å£°ãéåããããªãããšãããããã¡ããã®åªããã®å®ç§ãªäœçŸã§ãã
- ð§ å€éšAPIã¬ã¹ãªãææ°ã®åçãå€å¥ããžãã¯:
- ã¹ã¯ããŒã«ãDOMã®åçãªåæç»ã«ãã£ãŠéå»ã®åçãåèªã¿äžããããåé¡ã«å¯ŸããGemini APIã«äžåäŸåãããããŒã«ã«ãã£ãã·ã¥ãšDOMæ§é ã®åæã®ã¿ã§ãææ°ã®åçããå€å¥ããèªåŸçãªããžãã¯ã確ç«ããŸããã
- ããã¯ãå€éšãµãŒãã¹ã«äŸåããªãUserScriptã®èšèšææ³ãäœçŸãããç¥çã§ãšã¬ã¬ã³ããªè§£æ±ºçã§ãã
- ð ãŒãã³ã³ãã£ã°ã®ãã«ãAI察å¿ãšæ¡åŒµæ§ã®ç¢ºä¿:
- GeminiãChatGPTãGoogleæ€çŽ¢AIã¢ãŒããšãã£ãè€æ°ã®å¯Ÿè©±åAIã«å¯Ÿå¿ãã€ã€ãæ¬æãããã¿ãŒã®æ¿å
¥äœçœ®ãã»ã¯ã¬ã¿é
åã§ç®¡çããããšã§ãå°æ¥ã®ãµãŒãã¹è¿œå ã«ãæè»ã«å¯Ÿå¿ã§ãããæ¥µããŠæŽç·Žãããæ¡åŒµæ§ã®é«ãã¢ãŒããã¯ãã£ãèšèšããŸããã
ð è£è¶³æ
å ± / 管çè
ã¡ã¢
æ¬ãªããžããªã«ã¯ããããžã§ã¯ãã®æ©èœãšã¯çŽæ¥é¢ä¿ã®ãªãã管çã»éçšäžã®ç®çã§é
眮ãããŠãã以äžã®ãã¡ã€ã«ãå«ãŸããŠããŸãã
googlef0587da5f8f69fa8.html: Google Search Consoleã®æææš©ç¢ºèªã®ããã«å¿
é ãªãã¡ã€ã«ã§ãã
ð çŸç¶ã®èª²é¡ / ä»åŸã®ã¿ã¹ã¯
- äžæããåŸã«åæã«åçãåéããã
- VOICEVOXæ¥ç¶ãšã©ãŒæã«ããŒã¹ããã§ãŠããªã
- VOICEVOXãžã®ãã©ãŒã«ããã¯åçæã«ãã©ãŒã«ããã¯åçã®ããŒã¹ããã§ãŠãªã
- AIã¢ãŒã/X(Grok)ã§ã¯ã³ã¯ãªãã¯ããªããšç£èŠã«ãŒããæ¢ãŸã£ãŠããã
éçºè
(Author)
ããã (Neon)
Bluesky :https://bsky.app/profile/neon-ai.art
GitHub :https://github.com/neon-aiart
GitHub Pages :https://neon-aiart.github.io/neon-spitch-link/
Greasy Fork :https://greasyfork.org/ja/users/1494762
chichi-pui :https://www.chichi-pui.com/users/neon/
iromirai :https://iromirai.jp/creators/neon
DaysAI :https://www.days-ai.com/users/lxeJbaVeYBCUx11QXOee