Holomate 专注数字人技术开发

Holomate All in Digital Human

下面作品参加2024大湾区展示 The following works participate in the 2024 Greater Bay Area Exhibition

ButtonGPT 一按即问大按钮

嵌入式数字人问答方案:一按即问大按钮 ButtonGPT

Embedded GPT voice solution: ButtonGPT One-Click Q&A Button

特点:

Features:

  • 单按键操作,简单方便,手掌大小
  • 内置提示灯,操作状态一目了然
  • 无需手机,独立运行
  • 支持调整回答速度和人声类型
  • 串口输出问题和答案的字符串,实时状态同步
  • 包含半年的大模型云服务
  • 外壳上的 USB 口不能读取串口数据,如需读取需要打开壳子后自行连接电脑
  • 不推荐使用充电宝作为供电源,因为充电宝的供电检测影响持续供电
  • 因为流量有限,为了保证每个用户半年的流量,每个大按钮目前限制每天20个提问。用户可以在配置页面输入自己讯飞云账号,实现无限制数量的提问
  • 点击下载外壳相关文件 shell.zip
  • 点击下载原理图相关文件 sch.zip
  • Single-button operation, compact size, palm-sized
  • Built-in indicator light for operation status
  • No phone needed, operates independently
  • Adjustable response speed and voice style
  • Serial output for questions and answers, real-time status synchronization
  • The USB port on the shell cannot be used to read serial data; to access serial output, open the shell and connect it directly to a computer.
  • Powerband is not recommanded as power supplier for its power detecting problem
  • Due to limited cloud service, every button is currently restricted to 20 questions per day. Users can input their iFlytek Cloud account on the configuration page to achieve an unlimited number of questions.
  • Click here to download shell-related files (shell.zip)
  • Click here to download sch files (sch.zip)

使用方法:

Usage Instructions:

A. 开机上网配置

A. Initial Setup and Internet Configuration

  • 大按钮插电即开机。开机后,用手机或电脑搜索 ButtonGPT 的热点,密码是 “password” 登录。
  • 用手机电脑的浏览器,打开192.168.4.1的网址(ButtonGPT默认的地址)。
  • 选择并填入需要连接的 WiFi 热点和密码,稍等后拔电重新上电开机。
  • Plug in the ButtonGPT to power it on. Connect to the "ButtonGPT" hotspot from a phone or computer, using the password "password."
  • Use the browser on your mobile phone or computer to open the website at 192.168.4.1 (the default address for ButtonGPT).
  • Select and enter the WiFi network name and password, wait briefly, then reboot by unplugging and replugging.

B. 开始使用

B. Start Using

  • 开机后,等待壳子面板内的红色灯点亮(不是插电后侧面的红灯),表示已连上网络。
  • 按住大按钮,直到灯亮表示接受语音提问,输出问题。
  • 大按钮连接云端获取答案,红灯亮起后稍等片刻即可听到答案。
  • After powering on, wait for the red light inside the surface(not that one in the side) to indicate network connection.
  • Press and hold the button until the light indicates it is ready to accept voice input. Speak your question.
  • The button connects to the cloud; once it has an answer, the red light will turn on, and the speaker will provide the answer shortly.

C. 个性化配置大按钮

C. Customize ButtonGPT

  • 按住按钮并插入 USB 线,进入配置状态并语音播报 IP 地址。
  • 在同一 WiFi 网络中访问该 IP 地址加上 “/edit” 后缀,进入 config 文件配置页面。
  • 可调整 tts_speed(语速 0-10之间)和 tts_person(朗读人 男人1=5003,小孩1=5118,男人2=106,小孩2=110,小孩3=111,女人1=103,女人2=5)等参数。
  • 在网址后加上 “/setup” 后缀,可重新设置 WiFi 密码或刷机。也可以在网址后加上 “/reset” 后缀重启,删除当前wifi配置
  • Press and hold the button while inserting the USB cable to enter setup mode. The IP address will be announced.
  • In a browser, go to this IP address with “/edit” to access the config file for customization (ensure on same WiFi).
  • Adjust parameters such as tts_speed (reading speed, between 0-10) and tts_person (voice choice Man1=5003,Kid1=5118,Man2=106,Kid2=110,Kid3=111,Lady1=103,Lady2=5).
  • Add “/setup” at the end of the URL to reset WiFi or update firmware. Also can add “/reset” at the end of the URL to delete WiFi config
桌边数字人一体机

视觉对话方案:桌面数字人一体机

Human Video GPT solution: Desktop Digital Human Device

Holomate全息伴侣,全球首个桌面全息数字人一体机。Holomate平台将说话、有生命的全息角色带入现实。无需打字,支持AI语音对话。支持不同种族的虚拟角色,包括英语、西班牙语、法语、德语、意大利语、日语、普通话甚至广东话等。

Holomate Holographic Companion, the world's first desktop holographic digital human device. The Holomate platform brings speaking, life-like holographic characters into reality, supporting AI voice dialogue without typing. Virtual characters of various ethnicities are available, supporting languages such as English, Spanish, French, German, Italian, Japanese, Mandarin, and even Cantonese.

特点:

Features:

  • 真人或卡通人动作互动,提供沉浸式体验
  • 语音问答,支持大模型内容生成
  • 全球10多种语音支持,满足多语言需求
  • Interactive actions with real or digital humans for immersive experience
  • Voice Q&A with large model content support
  • Supports over 10 languages globally

硬件产品特性:

Hardware Features:

  • 屏幕分辨率:1920像素(水平)× 1080像素(垂直)以上
  • Screen resolution: 1920 pixels (horizontal) × 1080 pixels (vertical) or higher

软件产品特性:

Software Features:

  • 唤醒方式:语音唤醒,按键唤醒,屏幕触摸唤醒,单次唤醒连续对话,单次唤醒单次对话
  • 识别语言:中文普通话,英语,法语,西班牙语,德语,意大利语,日语,韩语,泰语,粤语(支持普通话混合)
  • 朗读语言:中文普通话,英语,法语,西班牙语,德语,意大利语,日语,韩语,泰语,粤语
  • 大语言模型:Openai3.5,Openai4.0,Spark3.0,Doubao
  • 数字人模型:支持VRM,PMX模型,真人形象
  • 数字人语音功能:实时唇部动作,可根据语句长短完成唇部动作。支持原地站立说话,摊手说话,侧头说话,手势说话
  • 数字人可呼唤动作:飞吻,转身,比心,摊手,蹲坐,起立,舞蹈,唱歌
  • 数字人信息功能:角色扮演,非实时知识类信息回答,实时天气报告,实时时事新闻朗读,企业自定义信息回答
  • Wake-up methods: voice wake-up, button wake-up, touch screen wake-up, single wake-up for continuous dialogue, single wake-up for single dialogue
  • Recognition languages: Mandarin, English, French, Spanish, German, Italian, Japanese, Korean, Thai, Cantonese (supports mixed Mandarin)
  • Speech synthesis languages: Mandarin, English, French, Spanish, German, Italian, Japanese, Korean, Thai, Cantonese
  • Large language models: OpenAI 3.5, OpenAI 4.0, Spark 3.0, Doubao
  • Digital human models: Supports VRM, PMX models, and real-life images
  • Digital human voice function: real-time lip motion matching sentence length, with standing, hand gestures, head tilts, and other speaking actions
  • Digital human interactive actions: blowing kisses, turning, heart hand gesture, open arms, sitting, standing up, dancing, singing
  • Digital human information features: role-playing, non-real-time knowledge-based answers, real-time weather updates, real-time news reading, and custom corporate information responses