talk-to-chatgpt

talk-to-chatgpt

语音交互增强ChatGPT使用体验

Talk-to-ChatGPT是Chrome和Edge浏览器扩展,为ChatGPT提供语音交互功能。通过语音识别和文字转语音技术,用户可与AI进行自然对话。扩展集成ElevenLabs API支持自定义语音,并提供语速、语音类型和语言等多项设置。这不仅增强了用户体验,还为老年人和残障人士提供了便利的AI交互方式。

Talk-to-ChatGPTChrome扩展语音交互AI助手可访问性Github开源项目

Talk-To-ChatGPT has been discontinued

It's been fun while it lasted, but for various reasons this extension will not be maintained anymore. You are welcome to create forks and improve upon it. I am unable to keep the extension up to date against OpenAI's changes, and besides, they are releasing desktop apps for MacOS and Windows soon, free of use, which will make this extension obsolete.


Talk-to-ChatGPT is a Google Chrome and Microsoft Edge extension that allows users to talk with the ChatGPT AI using their voice (speech recognition), and listen to the bot's answer with a voice (text-to-speech), rather than just by typing. With this tool, users can speak to the AI and receive spoken responses, making the interaction feel more natural and conversational. This allows you to have your own personal assistant in the style of Iron Man's Jarvis, which is fun, but it's also a great way to help the elderly and people with disabilities interact with ChatGPT. We now support ElevenLabs API integration, which means you can create your own voices for text-to-speech!

The extension can be downloaded from here:

After installing the extension, open or reload the ChatGPT page ( https://chatgpt.com/ ) and you should be seeing a 'Start' button on the top right corner of the page. After you click Start, you will be asked for permission to use your Microphone. This is required to enable voice recognition.

Talk-to-GPT Widget

Once started, Talk-to-ChatGPT displays a menu on the top right corner of the page where users can access settings (such as voice, language, and more), skip the current message, toggle voice recognition on or off, and toggle text-to-speech on or off.

The settings menu can be seen below. Settings are saved in a cookie and reloaded automatically each time you activate the script.

Talk-to-GPT Settings 1

Talk-to-GPT Settings 2

Demo V2.6 with ElevenLabs support: https://www.youtube.com/watch?v=gnijfWuenKA

How to install

Option 1: CHROME or EDGE STORE

Option 2: MANUAL INSTALL

If the extension is temporarily unavailable (this can happen when OpenAI make breaking changes), or if you want to install the latest updates before they are available on the Chrome/Edge web store, you can install the extension manually. Here is how you do it.

  1. Download the .zip file here: https://github.com/C-Nedelcu/talk-to-chatgpt/raw/main/chrome-extension/chrome-extension.zip (this link will always point to the latest version)
  2. Extract the .zip file in a folder somewhere
  3. Follow this tutorial to install the extension in Chrome/Edge in dev mode: https://webkul.com/blog/how-to-install-the-unpacked-extension-in-chrome/

FAQ

Q: Which web browsers are supported? A: This extension is designed for Google Chrome and Microsoft Edge, desktop version only. The extension will not work in any other web browser, especially not mobile browsers, because these browsers do not support the necessary APIs for speech recognition and speech synthesis.

Q: Can you make it speak faster or in a different voice or language? A: Yes, use the settings menu. You can select a variety of settings among which the speech rate, voice type, and language.

Q: What is the purpose of this project? A: Originally, it was mostly a fun proof of concept. This AI is mind-bogglingly intelligent and I had a deep desire to converse with it orally, to make it more interesting. Surely OpenAI themselves will make a proper voice-controlled version of ChatGPT in the future, at which point my project will be rendered useless. For now, it seems to be helping people with disabilities / visually impaired people, so I'm going to be actively working on the project for as long as I can, as a form of contribution to society.

Q: Is it safe to use? A: Yes, it's simple javascript code that will execute only in the context of the ChatGPT webpage. It doesn't request any particular permissions, and it is fully open source. As soon as you navigate away from ChatGPT, everything is cleared, except for the addon settings.

Q: Will it always work? A: it might not work indefinitely, and here's why. The code is based on the current HTML structure of the ChatGPT page. If OpenAI change the HTML code, this project will likely stop working. I will probably keep updating it to maintain compatibility, but I'm not sure I'll be doing that forever. If you want to contribute to the project you are more than welcome to submit your own changes through Github.

Q: I have an error or a problem... A: Feel free to update the javascript yourself and propose changes on Github, or simply report the issue if you aren't a programmer.

Q: Can I make changes to your code? A: Yes, feel free to make changes, and do whatever you want, commit, fork, just have fun.

Q: How do I know what languages are supported? A: this is entirely based on the web browser APIs (Google Chrome, Microsoft Edge), so you need to ask Google or Microsoft, as I cannot provide an up-to-date answer. I've only tested it with English, French, and Chinese. The languages in the settings menu are the same ones found on the Google and Edge demos.

Press coverage

Talk-to-ChatGPT has been receiving press coverage since its release. It is currently featured on the following sites:

Change log

Version 2.9.0 - December 3rd, 2023

  • New: added Azure Text-To-Speech (#210) Full credits: lout33
  • Updated: ElevenLabs - added eleven_english_sts_v2 and eleven_turbo_v2 models
  • Fixed: ElevenLabs - cloned voices can now be used with all models (#200)

Version 2.8.1 - October 9th, 2023

  • Fixed: conflict with OpenAI's new image upload functionality (#189)

Version 2.8.0 - September 30th, 2023

  • Updated: more compact widget (#186)
  • Updated: lot of code refactoring and cleanup (credit: PxPerfectMike)

Version 2.7.2 - September 23rd, 2023

  • New: added an option to prevent the text-to-speech engine from reading emojis (#166) Credit: PxPerfectMike
  • Updated: added a note in the voice command setting area to remind users that voice control only works when the microphone is active
  • Fixed: red bar indicating microphone activity didn't activate after a pause or complete stop

Version 2.7.1 - September 17th, 2023

  • Updated: initial position of the widget lowered a bit so that the ChatGPT 'Share Chat' icon is visible (#163)
  • Updated: in the ElevenLabs voice list, it is possible to select v1 or v2 multilanguage model
  • Fixed: monolingual voices in ElevenLabs didn't work in v2.7.0 due to a mistake in the API code

Version 2.7.0 - September 16th, 2023

  • New: browser warning on first use to better inform those who are not running supported browsers
  • Updated: switched to ElevenLabs v2 models for better text-to-speech quality (#168)
  • Fixed: CRITICAL - ElevenLabs text-to-speech stopped working due to a CSP update from OpenAI. All ElevenLabs TTS playback was migrated to 'offscreen' (#177, #176, #169...)
  • Fixed: prevented commas within numbers from breaking sentences (Credit: decfrr)

Version 2.6.1 - June 3rd, 2023

  • Fixed: major issue with ChatGPT page update preventing from sending the prompt (#127, #128, #129, #130, #131)
  • Fixed: saving settings wasn't working when the list of system text-to-speech voices was empty (#125)

Version 2.6.0 - April 29th, 2023

  • New: ElevenLabs API support (#70, #41, #34)
  • Updated: beep tone at the start instead of saying 'OK', so as to avoid consuming ElevenLabs quota
  • Fixed: stability improvements

Version 2.5.0 - April 25th, 2023

  • New: beep tones when pausing and unpausing (#83)
  • Updated: the pause word will work even when spoken twice instead of once, to tackle issue with the speech recognition API
  • Updated: settings menu divided into sections
  • Fixed: forbid punctuation in voice commands as they may break settings storage
  • Fixed: text-to-speech sometimes wouldn't work when 'skip code blocks' option was enabled due to jQuery issue

Version 2.4.0 - April 24th, 2023

  • New: the widget can be dragged around
  • Fixed: issue with jQuery being unable to read message contents and speak out loud

Version 2.3.0 - April 18th, 2023

  • New: option to skip code blocks while reading bot's response (#17)
  • Fixed: when speech recognition returns an empty sentence, this could cause the addon to disfunction (#72)

Version 2.2.0 - April 16th, 2023:

  • New: 'pause' vocal command was revamped. When you say the pause word, the addon will continue listening but won't send anything to ChatGPT. To resume normal functionality, say the pause word again, or click the Resume button
  • Updated: minor UI and text updates here and there
  • Bugfix: when the addon is started in the middle of a conversation with ChatGPT, the last message received was always spoken out loud. It won't be the case anymore.

Version 2.1.0 - April 15th, 2023:

  • New: UI was revamped. Credits to pixelsoda and Shaun James

Version 2.0.2 - April 15th, 2023:

  • Fixed: the addon will properly activate on every ChatGPT conversational pages

Version 2.0.1 - April 14th, 2023:

  • Fixed: the ChatGPT URL was changed by OpenAI, therefore disabling the addon

Version 2.0.0 - April 10th, 2023:

  • New: keyboard shortcuts to control the addon. The shortcuts are listed in the settings menu or appear when you move the mouse over the UI buttons
  • New: an option to avoid breaking down sentences with commas or other punctuation marks such as colons and semicolons
  • Fixed: vocal commands such as the stop and pause words wouldn't always work because the speech recognition would add a dot at the end

For older change logs, please check the commit messages: https://github.com/C-Nedelcu/talk-to-chatgpt/commits/main

Donations

Thanks for reading all the way down. Are you enjoying Talk-To-ChatGPT and want me to continue improving it? You can help by making a donation to the project. Please click the Donate button to proceed.

<img alt="paypal" height="80" src="https://edunext.com.sg/paypal.png" />

编辑推荐精选

Trae

Trae

字节跳动发布的AI编程神器IDE

Trae是一种自适应的集成开发环境(IDE),通过自动化和多元协作改变开发流程。利用Trae,团队能够更快速、精确地编写和部署代码,从而提高编程效率和项目交付速度。Trae具备上下文感知和代码自动完成功能,是提升开发效率的理想工具。

AI工具TraeAI IDE协作生产力转型热门
问小白

问小白

全能AI智能助手,随时解答生活与工作的多样问题

问小白,由元石科技研发的AI智能助手,快速准确地解答各种生活和工作问题,包括但不限于搜索、规划和社交互动,帮助用户在日常生活中提高效率,轻松管理个人事务。

热门AI助手AI对话AI工具聊天机器人
Transly

Transly

实时语音翻译/同声传译工具

Transly是一个多场景的AI大语言模型驱动的同声传译、专业翻译助手,它拥有超精准的音频识别翻译能力,几乎零延迟的使用体验和支持多国语言可以让你带它走遍全球,无论你是留学生、商务人士、韩剧美剧爱好者,还是出国游玩、多国会议、跨国追星等等,都可以满足你所有需要同传的场景需求,线上线下通用,扫除语言障碍,让全世界的语言交流不再有国界。

讯飞智文

讯飞智文

一键生成PPT和Word,让学习生活更轻松

讯飞智文是一个利用 AI 技术的项目,能够帮助用户生成 PPT 以及各类文档。无论是商业领域的市场分析报告、年度目标制定,还是学生群体的职业生涯规划、实习避坑指南,亦或是活动策划、旅游攻略等内容,它都能提供支持,帮助用户精准表达,轻松呈现各种信息。

AI办公办公工具AI工具讯飞智文AI在线生成PPTAI撰写助手多语种文档生成AI自动配图热门
讯飞星火

讯飞星火

深度推理能力全新升级,全面对标OpenAI o1

科大讯飞的星火大模型,支持语言理解、知识问答和文本创作等多功能,适用于多种文件和业务场景,提升办公和日常生活的效率。讯飞星火是一个提供丰富智能服务的平台,涵盖科技资讯、图像创作、写作辅助、编程解答、科研文献解读等功能,能为不同需求的用户提供便捷高效的帮助,助力用户轻松获取信息、解决问题,满足多样化使用场景。

热门AI开发模型训练AI工具讯飞星火大模型智能问答内容创作多语种支持智慧生活
Spark-TTS

Spark-TTS

一种基于大语言模型的高效单流解耦语音令牌文本到语音合成模型

Spark-TTS 是一个基于 PyTorch 的开源文本到语音合成项目,由多个知名机构联合参与。该项目提供了高效的 LLM(大语言模型)驱动的语音合成方案,支持语音克隆和语音创建功能,可通过命令行界面(CLI)和 Web UI 两种方式使用。用户可以根据需求调整语音的性别、音高、速度等参数,生成高质量的语音。该项目适用于多种场景,如有声读物制作、智能语音助手开发等。

咔片PPT

咔片PPT

AI助力,做PPT更简单!

咔片是一款轻量化在线演示设计工具,借助 AI 技术,实现从内容生成到智能设计的一站式 PPT 制作服务。支持多种文档格式导入生成 PPT,提供海量模板、智能美化、素材替换等功能,适用于销售、教师、学生等各类人群,能高效制作出高品质 PPT,满足不同场景演示需求。

讯飞绘文

讯飞绘文

选题、配图、成文,一站式创作,让内容运营更高效

讯飞绘文,一个AI集成平台,支持写作、选题、配图、排版和发布。高效生成适用于各类媒体的定制内容,加速品牌传播,提升内容营销效果。

热门AI辅助写作AI工具讯飞绘文内容运营AI创作个性化文章多平台分发AI助手
材料星

材料星

专业的AI公文写作平台,公文写作神器

AI 材料星,专业的 AI 公文写作辅助平台,为体制内工作人员提供高效的公文写作解决方案。拥有海量公文文库、9 大核心 AI 功能,支持 30 + 文稿类型生成,助力快速完成领导讲话、工作总结、述职报告等材料,提升办公效率,是体制打工人的得力写作神器。

openai-agents-python

openai-agents-python

OpenAI Agents SDK,助力开发者便捷使用 OpenAI 相关功能。

openai-agents-python 是 OpenAI 推出的一款强大 Python SDK,它为开发者提供了与 OpenAI 模型交互的高效工具,支持工具调用、结果处理、追踪等功能,涵盖多种应用场景,如研究助手、财务研究等,能显著提升开发效率,让开发者更轻松地利用 OpenAI 的技术优势。

下拉加载更多