Request to Skip Image URL in TTS Responses for Better Child Interaction #1210

New Issue

GiteaMirror · 2025-11-11T14:40:13-06:00

GiteaMirror commented

2025-11-11 14:40:13 -06:00

Originally created by @snakeying on GitHub (Jun 11, 2024).

Firstly, I would like to extend my heartfelt gratitude to the development team and all contributors for bringing us such an amazing project. This is undoubtedly the best open-source UI I have ever experienced on GitHub.

Recently, while using the project, I encountered a minor issue. I have two five-year-old children who are very interested in interacting with AI. Since they have not yet mastered sufficient typing skills, I set up Speech-to-Text (STT) and Text-to-Speech (TTS) functions for them, and everything seemed to work perfectly. To help them better understand the AI interaction, I used prompts that included both text and images, such as:

Please remember for the following interaction, your conversational partner is a five-year-old child:

Use a tone that is understandable and interesting to young children.
Must include images.
Be humorous and friendly.

As a result, an image is generated with each AI interaction. After enabling the auto-playback response in the audio settings, the TTS reads the image URL (https://xxxxxxxxxx) before proceeding to the text of the interaction. I would like to inquire if it is possible to add a filtering option to skip the reading of the image URL?

Thank you for your hard work, and I look forward to your response.

Originally created by @snakeying on GitHub (Jun 11, 2024). Firstly, I would like to extend my heartfelt gratitude to the development team and all contributors for bringing us such an amazing project. This is undoubtedly the best open-source UI I have ever experienced on GitHub. Recently, while using the project, I encountered a minor issue. I have two five-year-old children who are very interested in interacting with AI. Since they have not yet mastered sufficient typing skills, I set up Speech-to-Text (STT) and Text-to-Speech (TTS) functions for them, and everything seemed to work perfectly. To help them better understand the AI interaction, I used prompts that included both text and images, such as: Please remember for the following interaction, your conversational partner is a five-year-old child: 1. Use a tone that is understandable and interesting to young children. 2. Must include images. 3. Be humorous and friendly. As a result, an image is generated with each AI interaction. After enabling the auto-playback response in the audio settings, the TTS reads the image URL (https://xxxxxxxxxx) before proceeding to the text of the interaction. I would like to inquire if it is possible to add a filtering option to skip the reading of the image URL? Thank you for your hard work, and I look forward to your response.

GiteaMirror closed this issue

2025-11-11 14:40:13 -06:00

GiteaMirror commented

2025-11-11 14:40:14 -06:00

@justinh-rahb commented on GitHub (Jun 11, 2024):

That would be dependent on the TTS API that you are using, afaik the OpenAI one filters out that sort of content automatically.. at least I've never had it try to read out a code block for example.

@justinh-rahb commented on GitHub (Jun 11, 2024): That would be dependent on the TTS API that you are using, afaik the OpenAI one filters out that sort of content automatically.. at least I've never had it try to read out a code block for example.

GiteaMirror commented

2025-11-11 14:40:14 -06:00

@snakeying commented on GitHub (Jun 11, 2024):

I used the “default” option

@snakeying commented on GitHub (Jun 11, 2024): I used the “default” option

GiteaMirror commented

2025-11-11 14:40:15 -06:00

@justinh-rahb commented on GitHub (Jun 11, 2024):

Yes, the Web Speech API provided by your browser/OS is responsible in that case. They're pretty unsophisticated.

@justinh-rahb commented on GitHub (Jun 11, 2024): Yes, the Web Speech API provided by your browser/OS is responsible in that case. They're pretty unsophisticated.

GiteaMirror commented

2025-11-11 14:40:15 -06:00

@snakeying commented on GitHub (Jun 11, 2024):

Well, I just try the openai TTS and it works like a champ...

Thx, and agian, keep up the great work~~ ^^V

@snakeying commented on GitHub (Jun 11, 2024): Well, I just try the openai TTS and it works like a champ... Thx, and agian, keep up the great work~~ ^^V

GiteaMirror referenced this issue

2026-04-19 19:17:59 -05:00

[GH-ISSUE #1210] add RAG: cpu/cuda/mps as a config variable rather than hardcode #12395

GiteaMirror referenced this issue

2026-04-25 02:41:04 -05:00

[GH-ISSUE #1210] add RAG: cpu/cuda/mps as a config variable rather than hardcode #27923

GiteaMirror referenced this issue

2026-05-05 11:52:22 -05:00

[GH-ISSUE #1210] add RAG: cpu/cuda/mps as a config variable rather than hardcode #51060

GiteaMirror referenced this issue

2026-05-12 23:17:27 -05:00

[GH-ISSUE #1210] add RAG: cpu/cuda/mps as a config variable rather than hardcode #66999