What message types does Telegram Customer Service Translator support? Text-based, media description and manual processing FAQ

When your cross-border team uses Telegram Bot to receive customers from different countries, language barriers are often the first threshold to cross. The automatic translation function seems to be a universal antidote, but many operators are confused in actual use: **Why can’t the text in the picture be automatically translated? Can voice messages be converted directly into text? **

This article takes the automatic translation function of TG-Staff as an example to break down the message type boundaries supported by Telegram customer service translator in detail, focusing on the automatic translation capabilities of text messages, the “explanation rules” of media messages, and how agents cooperate with manual processing in actual work. Whether you are evaluating customer service tools or already using TG-Staff, this article can help you more accurately understand the applicable scenarios of the translation function.

---

Why does Telegram customer service translator mainly focus on text messages?

The bottom layer of automatic translation technology is natural language processing (NLP), and the input object it processes is text string. Media files such as pictures, videos, and voices are essentially binary data or unstructured information and cannot be directly parsed and output in another language by the translation engine.

Core capabilities of automatic translation: real-time conversion of text messages

When a customer sends a plain text message (such as “What is the delivery time for order #123?”), TG-Staff’s automatic translation function will:

1. Detect source language: The system automatically identifies the message language (such as English).
2. Match target language: Based on the language preference set by the agent in the web console, or the target language (such as Chinese) in the session context.
3. Real-time translation and display: On the agent’s chat interface, the translated text is displayed above or below the original text; when the agent sends a reply, the system also translates the agent’s native text into the customer’s language and then sends it to the customer.

The entire process is completed in 1–2 seconds, and the agent sees a bilingual comparison and the customer sees their own language.

Why can’t media messages (pictures, videos, documents) be automatically translated?

• Image: The translation engine cannot recognize the text in the image (OCR is an independent technology, TG-Staff currently does not have built-in OCR).
• Video/Speech: There is no audio-to-text (ASR) module, and translatable text cannot be extracted.
• File (PDF/Word/Excel): The file content cannot be parsed directly by the translator unless the agent manually opens the file and extracts the text.

Therefore, when a customer sends a screenshot or a piece of voice, the system will not automatically translate the media itself. Instead, it will mark “media message” in the conversation record and wait for the agent to process it.

---

List of text message types supported by TG-Staff translation function

Although media messages require manual cooperation, TG-Staff’s automatic translation supports text messages very comprehensively, covering more than 90% of the information in customer service communications.

Plain text and rich text messages

Message type	Whether to support automatic translation	Description
Plain text	✅ Yes	The most basic message type, no format, real-time translation
Markdown format text	✅ Yes	Such as **加粗**, *斜体*, [链接](url), retain the format after translation
HTML format text	✅ Yes	Such as <b>加粗</b>, the tag structure is retained after translation
Bot command reply	✅ Yes	Bot automatic reply text will also be translated
System prompt message	❌ No	For example, “the session has been transferred”, etc., the system will not translate it to avoid ambiguity

Actual scenario: The customer sends an order query message in Markdown format in English:

Hi, can you check the status of **order #456**?  
My email is [[email protected]](https://app.tg-staff.com/)

After TG-Staff automatically translates, the Chinese version seen by agents will retain the bold and link format, making it easier for agents to click directly.

Trigger conditions and language detection mechanism for automatic translation

The translation function of TG-Staff is enabled by default (standard version and above packages), and the triggering conditions are very simple:

• Agent side: Agents set their own language preferences (such as “Chinese”) in the web console. When a customer sends a non-Chinese message, the system automatically translates it into Chinese and displays it to the agent.
• Customer side: When the agent sends a reply, the system translates the reply into the customer’s language based on the customer’s language in the conversation history (or the language manually specified by the agent) before sending it.

Language detection is automatically completed by the translation engine, which supports mutual translation in more than 100 languages ​​(the specific language list is subject to the official TG-Staff documentation). If the detection fails (for example, the message is too short or contains a lot of spelling errors), the system will display the original text by default and prompt “Language detection failed, please select manually.”

---

Translation and processing instructions for media messages such as pictures, videos, voices, files, etc.

When customers send media messages, TG-Staff will not automatically translate the media content, but the system will assist agents in the following ways:

1. Conversation Record Mark: In the chat list, media messages will display icons (such as 📷 pictures, 🎤 voice) and be marked “Media Message”.
2. Agent enters description manually: After the agent can view the media content, he can write a text description in the input box and then send it. This description text will be automatically translated to the customer.
3. History traceability: All media messages are retained in the session history, and agents can review them at any time.

Best practices for agents to manually handle media messages

The following process applies to TG-Staff or any customer service system that does not support automatic media translation.

1. Picture Category:
   • If the picture contains text (such as product screenshots, invoices), the agent manually enters “The text in the picture is: XXX, which means YYY”, and the system automatically translates this description.
   • If the picture is a product photo, the agent can enter “Thank you for sending the picture. I have seen the appearance of the product. What information do you want to know?”
2. Voice message:
   • Voice messages cannot be translated directly. The agent can reply “I received your voice message, please describe your problem in text, or I will contact you later to communicate via text.”
   • If the team has an AI transcription tool, it can be transcribed into text first, then pasted into TG-Staff and sent (see next section).
3. File (PDF/Word):
   • The agent downloads the file and extracts key information, summarizes it in his own words and sends a text reply.

Transcription and translation of voice messages (requires third-party cooperation)

TG-Staff itself does not provide speech-to-text (ASR) functionality. But teams can compensate by:

• Use third-party speech transcription APIs (such as Google Cloud Speech-to-Text, Whisper) to convert speech files into text.
• Paste the transcribed text into the chat input box of TG-Staff, and the system will automatically translate it and send it to the customer.

Note: This solution requires additional tools and manual operations, and is suitable for teams with frequent voice messages. If there are very few voice messages, directly asking customers to communicate via text is a more efficient option.

---

What does “Description Rules” in the Customer Service Translator mean?

When many users see “Media messages require manual processing” or “Description rules” in the TG-Staff document, they misunderstand that the system will automatically generate instructions for the media. In fact, the statement rule means:

When an agent receives a media message, the system will not automatically parse the media content. The agent needs to actively view the media and then manually enter the text description before the description text will be translated by the system and sent to the customer.

For example:

• ❌ Misunderstanding: The customer sent a Japanese menu picture, and the system automatically translated it into Chinese text and displayed it to the agent.
• ✅ Correct process: After the agent sees the picture, he manually enters “This picture is a Japanese menu, the main content is…”, and the system translates this Chinese description into the customer’s native language (such as English) and sends it back.

This rule may seem to increase manual workload, but it ensures the accuracy of translation - image descriptions automatically generated by machines are likely to be wrong or miss key information, while manually entered content is more reliable.

---

How can cross-border customer service teams use translators to improve communication efficiency?

After understanding the capabilities of the translator, the cross-border team can optimize the workflow in a targeted manner:

Scenario 1: Multi-language customers access at the same time

• Problem: English, Japanese, and Spanish customers send messages at the same time, and the agent needs to switch to reply in different languages.
• Solution: TG-Staff’s automatic translation allows each agent to respond to all customers in the language they are most familiar with (such as Chinese). The system automatically translates Chinese replies into each customer’s language. Agents do not need to learn multiple foreign languages.

Scenario 2: Text messages are given priority, media messages go through manual process

• Best Practice: In the Bot welcome message, guide customers to first describe the problem in words (“Please describe your needs in words so that we can serve you faster”), and reduce the proportion of media messages.
• Handling Media: Arrange an agent to handle media messages exclusively, or specify that media messages must be manually responded to by an agent with text within 30 minutes.

Scenario 3: Combining diversion links and automatic translation

• Process: Advertising traffic → Customer clicks the diversion link → Bot automatically replies (including text guidance) → Customer sends text → Automatic translation → Agent replies in native language.
• Effect: In the entire link, the translation of text messages is completely automated, and only media messages require manual intervention. For teams such as cross-border e-commerce and overseas gaming, this model can significantly lower the customer service threshold.

Scenario 4: Combining content risk control and translation

• Professional Edition Users: Before the agent sends the translated message, the content risk control function will detect risky words (such as wallet addresses) to prevent the leakage of illegal information. Translation and risk control run simultaneously without affecting efficiency.

---

Frequently Asked Questions (FAQ)

Q: What languages does TG-Staff’s automatic translation support?

Answer: Both the TG-Staff standard version and the professional version support mutual translation in more than 100 languages, including English, Chinese, Japanese, Korean, Spanish, French, German, Arabic, Portuguese and other mainstream languages. The specific language list can be viewed in the official documentation. The translation engine is powered by AI (Standard) or switchable Google/DeepL professional translation (Pro), with consistent language coverage but higher quality translations in Pro.

Question: If a customer sends a picture, will the translator translate the text in the picture?

Answer: No. TG-Staff’s translator only processes text messages and cannot recognize text within images. The agent needs to manually view the picture and enter a text description, and then the description text will be translated. The system does not automatically generate descriptions or extract text for images.

Question: Can voice messages be automatically translated into text?

Answer: Currently, TG-Staff does not support the speech-to-text (ASR) function. Voice messages cannot be translated automatically. The agent can ask the customer to communicate by text instead, or use a third-party speech transcription tool to convert the speech into text, and then paste it into TG-Staff and send it.

Q: Will the translated message show the original text? Can agents see both languages?

Answer: Yes. In TG-Staff’s web console chat interface, agents see a bilingual comparison: the original text is at the top and the translated text is below (or vice versa, depending on the configuration). This way the agent can check whether the translation is accurate and manually correct it if necessary. The client only sees the translated language, not the original text.

Q: Can I use the translation function during the free trial?

Answer: Yes. TG-Staff offers a 3-day free trial, during which all features of the Standard and Professional versions are available, including automatic translation. After the trial expires, the translation function will be suspended along with the package status and will be automatically restored after renewal. It is recommended to test the accuracy and response speed of the translation during the trial period to confirm whether it meets the needs of the team.

---

Start using TG-Staff’s automatic translation function

If your cross-border customer service team is looking for a Telegram customer service tool that can automatically translate text messages and support manual processing of media messages, you might as well start with TG-Staff’s free trial.

• Register for trial: Visit https://app.tg-staff.com/ to create an account and experience the full functionality within 3 days.
• Check the documentation: Learn more about translation function configuration and language list → https://docs.tg-staff.com/
• Contact Customer Service: If you encounter problems, you can directly contact the official Bot @tgstaff_robot for real-time help.

Only by mastering the capabilities of the translator can the customer service process be more efficient. Automatic translation of text messages is the core of efficiency, and manual processing of media messages is a necessary supplement. The cooperation of the two is the best practice for cross-border Telegram customer service.