For Windows Version 1.0.1 One-key package. You can get Yachiyo from Tsukuyomi Space Room back to have a voice. For the first time, it is necessary to download the decompression, then turn on the starter, wait for it to be ready and then enter the site it opens.

If you can already talk to Yachiyo in Room, start with the first step. A one-key package automatically fills out the voice arguments, you do not need to learn the code or buy another API key for the voice service.

What do you want before you start?

Prepares a computer to run Windows, headphones or speakerboxes to play sound, and browsers to open Tsukuyomi Space. This course runs a one-key package, opens a website and listens to sound on the same computer.

This one-key division is simple:The chat model answers, and a key pack reads the text into sound. If Room is not able to reply to the text, please finish " What if the text is not configured yet?" The website account and chat provider account are different; login does not automatically match chat models.

Networks are needed to download and open the website. The voice model is running on your computer; the Japanese version of website chat and pre-reading uses your configured chat model. When using a cloud chat service, these calls are charged according to the rules of the service provider.

Step one: Download the full one.

  1. ClickYachio voice one-key download chain Answer.I don't know. If the disk requests extraction code, enter 66Mj。
  2. Double click open Word synthesis one. folder.
  3. Found it. YachiyoTTS-1.0.1-Windows-AMD-NVIDIA - Default Browser.zipI don't know. Share page shows it about 7.5GI don't know. Tick the box in front of it and select only this complete ZIP file.
  4. If Quake hints to log in, log your own Quake account, then use the download portal available on the web. If you need to start. Save to Web, save and then download it on your own disk; if the page requires a client, click the quark page hint. Web login is only for documents.
  5. Wait till the download is finished. The next step is taken after the browser or download software shows " finished " .

In the same directory .zip.sha256 The document is approximately 131B, which is used to verify whether the downloaded file is complete and cannot replace the complete package of 7.5G. Normal users download first on the course .zip Yeah.

A complete ZIP package of about 7.5G should be selected from the disk

Signs of completion of this level: Your computer is already complete. .zip File, download is over.

Step two, compress the package into a folder.

The "pressure" is to remove all the documents in the compressor bag. Double-clicking the program directly in the compressed package window, and it's not the same thing to run when it's depressed.

  1. Found the download. .zip Documentation. The browser download list usually has a " Show in the folder " or similar entry.
  2. Right mouse click it, choose Extract AllI don't know. If you use other decompression software, you can also select the " unpressure to folder " function of the software.
  3. Select an easy location, for example D:\YachiyoTTSI don't know. This is an example of a place where you can choose a space on your computer without a D drive.
  4. Click to depress and wait for progress to be completed. The documents are larger and the waiting time is related to computer and hard drive speed.
  5. Opens a folder after depressure. All follow-up started from here.

Hard disks need to accommodate both downloaded ZIP and settled files; only about 7.5G space is not enough. When there is a “space shortage”, the disk is released in sufficient space.

Signs of completion of this level: The decompression is over, and you opened a normal folder in which you can see a one-key package.

Step three, turn on the starter. Okay.

  1. In the home folder after depression, find in the package description Starter or Start ScriptTwo clicks open. The specific file name for the start-up entry is indicated in the package.
  2. If the starter provides the hardware or mode of operation options, select the corresponding method according to the package description.
  3. Waiting for voice model to load. Keep the window open while loading; if you have a clear error hint, click the fail-check at the back.
  4. After the model is ready, the web page is opened using a one-key package. 1.0.1 Tsukuyomi Space Room will be opened through the Windows default browser.
  5. If the browser asks whether the site is allowed to access its own services or local networks, it will not be able to access the Internet.After confirming that the current site is Tsukuyomi Space, select permissionI don't know. This allows the web page to connect to the voice program on your computer.

Keep the starter and the service window that it opens running. These windows are responsible for providing sound, and when the service is closed, the web page cannot continue to be called.

The link opened by the starter gives this connection to the website. The site was then restored to the normal Room address, and the site has saved voice configuration to the current browser.

Signs of completion of this level: The model is ready, and Tsukuyomi Space Room is shown in the browser that the starter opens.

Step four, see if the site is automatically connected.

  1. Opened in the browser.Room Settings。
  2. Click on the left Voice and SpeechI don't know. If the window is narrower, click first. All SettingsSelect again. Voice and Speech。
  3. Inspection Enable speech synthesis The switch has already been turned on.
  4. Check for voice services GPT-SoVITS (Yachiyo one key pack / machine)。
  5. See if the page shows:Yachiyo is automatically connected and can read directly. Please keep the starter running.

If these contents already appear, the parameters are automatically imported and saved.Advanced voice settings remain as they are. API keys and acoustic numbers are not required.

Example of voice settings after automatic access

The above figure is a test screenshot of the automated access process to the website, showing web-page tips and button locations.

If the switch is still closed or the voice service is the other service selected previously, please return to the one-key package and use it again to open the website. Separately enter Room from the collection and do not retrieve the connection information for this startup.

Signs of completion of this level: Switches are on, services are selected, and there is a one-key package of automatic connection tips on the page. This hint indicates that the parameter is imported; the next test is used to confirm that the service does make a sound.

Step five, give it a try.

  1. Confirms that headphones or speaker boxes are connected, Windows volume is open, and the browser tab is not silent.
  2. Yes. Voice and Speech Page Click Save and testI don't know. The configuration was automatically saved, and this step is to test the sound.
  3. Waits for the synthesis to be completed, so do not repeat the click.
  4. Confirms the current Tsukuyomi Space page as the back of the first popup access to this machine AllowTry again if necessary.
  5. When the site starts to play successfully, it will appear. This is a direct request for GPT-SoviTS and start playing. The tip.
  6. When you actually hear the sound, close the window and click. Go back to your room.。

The web page displays success but no sound on the headphone, continues to check the system volume, browser silence and voice output until you actually hear it.

Signs of completion of this level: You can hear the speech.

Step six, let Yachiyo read the chat.

  1. BackPrivate Room Room。
  2. Write a short sentence in the chat input box, for example:Hello. Say hello to me in one sentence.
  3. Click to send, wait for Yachiyo's text to appear.
  4. Click under this response. Play Voice。
  5. button display Loading while waiting; displaying Playing After hearing voices. It is normal that buttons cannot be clicked again while loading.

The current GPT-SoviTS chat will be read in Japanese first. So you may see Chinese answers, hear Japanese. This step uses the chat model you configured in Room.

If you set the page to hear sound, but read error in the chat, see if the error refers to chat models, LLM or Japanese Transports. This should be checked. Chat Model Configure, voice packs are not necessarily problematic.

Signs of completion of this level: Room can get a text response and actually hear the sound by clicking on the "play voice" response.

What if you haven't configured a text chat yet?

Those who have been able to talk regularly can skip this section. When you first use Room, press the following:

  1. OpenRoom Settings, click Chat Model。
  2. If you have an API key for a chat service, select the corresponding service provider. The API key is a sequence of characters that allow a web page to call the service. It cannot fill in a web login password or chat app password.
  3. If you have no key, register or log in on the official platform of the provider of your choice. API Key / API Key The page takes its own key and confirms that the account number has an API amount available.
  4. The person who pasted the key to the website API Key Input box, in Model Enter the full model name provided by the service provider. When using the built-in service provider option, the website usually fills in the corresponding interface address.
  5. Click Test ConnectionI don't know. After testing, click Save and return to the room。
  6. Sends a short message confirming that you can get a text response and then continues to test the chat.

The interface address, key and model must come from the same service provider. People who already have a local Ollama chat model can follow their own configuration; this voice one-key package does not in itself install chat models for you.

The key is saved in the current browser. Do not send the key to the square, comment or chat message; also cover the key and the full address with the link document before showing the wrong screenshot.

We'll use it the next day.

Downloads and decompressions are usually done only once. And every time that follows, in this order:

  1. Opens a folder after depressure.
  2. Turn on the starter and wait for the model to be ready.
  3. Use the starter to open the Tsukuyomi Space page.
  4. Keep starter running, click on reply Play Voice。

Each start-up is replaced by a connection certificate, and the web site that the starter opens automatically updates it.Reopen the site through the starter after restarting a key bag, the last connection can be avoided.

After changing browsers, removing website data or changing computers, you need to re-open the site through the starter in the new browser. The key to the chat model may also need to be refilled in the new browser.

If it doesn't work, it's a phenomenon.

Only one small file after downloading.

Check file names first..zip.sha256 Validation file about 131B; complete package about 7.5G .zipI don't know. Go back to the web to select the complete ZIP download.

Unpressure Times files damaged or downloaded incompletely

Confirms that the download is over and checks if the download software is wrong. Redownload full ZIP, complete decompression. Do not mix the scattered documents in the old and new versions.

No start after double click or window disappears.

Confirms that the pressure has been fully decompressed and that the entry in the accompanying description is activated. Read first the error or log shown in the one-key package; if there is a hardware, drive or relying problem, treat it as specified in the package or provide the error to the publisher.

There is no automatic connection hint on the page

Keep the package running and re-open the site through the starter. Confirms that your viewing of the set browser is the browser that the starter opens. If the site hint is unable to save the connection, the site is allowed to be stored with a browser and opened from the starter again.

GPT-SoviTS

Check if the starter and service are still in operation, if the model is ready, and if the browser allows Tsukuyomi Space access to its own service. Make sure one key bag and web page are on the same computer.

If you have previously refused to request access to this machine, open the website access to the site or browser next to the browser address bar to find access to the local network; the entry name is shown on the browser. Allow to return to the web site for an audition.

401 or connection error

This indicates that the request was not certified by the service. Reactivate the one-key package and re-open the site through the starter and try again. Do not manually delete the connection information that is automatically taken after the API address.

The audition succeeded, but the playback failed.

Let's make sure that the text chats are normal and then check if there's a mistake. LLM, Japanese Transports, Amounts, Keys or ModelsI don't know. Chats need to be completed first, and then you can try to fix the chat model by hint.

Show playing but no sound

Checks if headphones or speaker connections, system volume, browser tabs are silent, and Windows is exporting the sound to another device. We'll use the computer to play a sound you know, make sure the device is soundable, and then go back to the site.

First synthesis waits longer.

The model operates in this machine at speeds associated with hardware, model loading status and text length. Test the short sentences first, and then the next time after the end of the current task; there is no fixed "all computers can finish in seconds".

Can you connect your cell phone directly to the computer's first key bag?

Auto-access to the same Windows computer. Here. localhost It means "the current device itself": it means a cell phone on a cell phone and does not automatically connect to a computer. Cell phones or other equipment connections require a separate and accessible service, which is not included in the curriculum.

Last check before use

  • The one-key package has been activated and the voice model is ready.
  • The starter and voice service windows are still running.
  • The site was opened from this start-up one-key package.
  • Sets the page already displays a one-key package of automatic connection tips.
  • Click Save and Test to actually hear the sound.
  • Room can reply to text normally, and click " Play Voice" can hear reply.

When the first course is completed, Journal simply remembers: turn on the starter, and when it is ready, use the site it opens and keep the service running.