Change Logs
2024-9-1 Server@2.0.0 Client@2.0.0
🔥🔥🔥 A new member has been added to the ecosystem: ESP-AI Developer Platform
🔥🔥🔥 Website: https://dev.espai.fun
🔥🔥🔥 Features in the Developer Platform: provides free services, visual configuration services, various online tools, and more...
Server
- 🔥 Added: When errors are caught, they will be fed back to the client, allowing clients to reference an error table for relevant error prompts.
- 🔥 Added: Built-in Volcano Engine
LLM
(compatible with Doupao and others) - 🔥 Added: Each user under the service can have their own independent set of configurations.
- 🔥 Optimized: The
TTS
task mechanism now supports incremental character conversion, making conversations smoother and faster while addressing concerns over TTS service costs. - 🔥 Optimized: During audio playback, LLM conversation history can be read, and support for controlling start times is included.
- 🆕 Added:
connected_reply
configuration, which allows changing the prompt message after service connection. - 🆕 Added: Output all configuration items upon service startup, making it easy for users to clearly know what services they are using.
- 🆕 Added: Exception handling to prevent services from crashing directly when not running via methods such as
pm2
. - 🤔 Adjusted: Removed the
api_key
configuration, replaced byiat_config
,llm_config
,tts_config
(as putting everything inapi_key
was confusing). - 🤔 Adjusted: Changed the default service port to
8088
(since there's a high chance of conflict with8080
). - 🤔 Adjusted: Other configuration parts, details can be found in the documentation.
- 💄 Optimized: Switched to a better prompt sound before starting a session.
- 🐞 Fixed: Issue where sentence breaks would repeat the previous sentence.
Installation
npm i esp-ai@1.x.x
Docker Installation:
docker exec -it esp-ai-server npm i esp-ai@1.x.x
Client
- 🔥 Added: Support for pairing with the
ESP-AI Developer Platform
for a super simple AI dialogue solution for individuals and enterprises. Simply enter anapi_key
on the network configuration page. - 🔥 Added:
.onError
unified error capture callback, allowing developers to decide how to notify users based on error codes. - 🔥 Added:
.onConnectedWifiCb
to monitor successful device connection towifi
, where the IP address can be obtained directly as a parameter. - 🔥 Added: Network configuration settings page, which supports customization. In addition to
wifi
information, it also provides additional fields for storing your business data. - 🔥 Added: Ability to interrupt conversations, by calling
Xiaoming Tongxue
to restart the session. - 🔥 Added: Multiple built-in wake-up solutions:
Pin level wake-up
,Tianwen module wake-up
,Serial string wake-up
, all ready to use with direct configuration. - 🔥 Added:
.onNetStatus
unified status callback, where the status of the device connecting to wifi or the service can be uniformly obtained. - 🔥 Added:
.onAPInfo
callback when the device opens a hotspot. You should prompt the user to open the configuration webpage here, and you can get all the hotspot information as parameters. - 🆕 Added:
setWifiConfig
for manually settingwifi
information. - 🆕 Added: Each device generates a UUID during initialization, which can be retrieved as needed, similar to a
sn
code. - 🤔 Adjusted: Configuration order adjusted, unnecessary parameters are placed at the end and can be left unconfigured.
- 🤔 Adjusting the built-in voice wake-up scheme is temporarily delayed until the next 3 or so versions are released (because precision optimization is too time-consuming, it needs to be upgraded separately).
- 💄 Optimized: Improved console output.
- 💄 Optimized: Refactored client-side code.
- 🐞 Fixed: Errors when uploading code in certain environments due to undefined
LED_BUILTIN
. - 🐞 Fixed: Various other issues...
Installation
Please download the latest version of the client from the repository release page: https://github.com/wangzongming/esp-ai/releases
The file to download is named esp-ai-xxx.zip
.
2024-8-1 Server@1.18.7 Client@1.4.2
Server
- 🆕 新增 网络延时输出
- 🤔 调整 服务启动时输出全部的本地IP地址
- 💄 优化 控制台部分文字颜色调整
- 🐞 修复 在音频数据过大时(如播放歌曲时),服务会异常断开连接
Install
npm i esp-ai@1.18.7
docker Install
docker exec -it esp-ai-server npm i esp-ai@1.18.7
Client
- 🆕 新增 网络延时输出
- 🐞 修复 在音频数据过大时(如播放歌曲时),服务会异常断开连接
Install
2024-7-30 Server@1.15.6 Client@1.3.1
Server
- 🆕 新增 鉴权配置,用户每次请求接口时都可进行一次鉴权
Install
npm i esp-ai@1.15.6
Client
- 🆕 新增 连接服务时可传递参数,用于服务端鉴权等。
Install
2024-7-23 Server@1.14.5 Client@1.2.1
Server
- 🐞 修复 播放
http
地址放入音乐会报错
Install
npm i esp-ai@1.14.5
Client
- 🐞 修复 一些其他问题
Install
2024-7-22 Server@1.14.4 Client@1.2.0
Server
- 🐞 修复 讯飞llm服务返回报错后框架未将错误输出
Install
npm i esp-ai@1.14.4
Client
- 💄 优化 稳定性提示
Install
2024-07-21
Server v1.14.3
- 🆕 Added the music player function
- 🐞 fixed no longer waking up after the first wake up
Client v1.2.0
- 💄 Stability optimization Tips
2024-07-20
Server v1.13.2
- 🆕 Added callback after speech recognition is complete.
- 🆕 Added vad_eos speech recognition silence time configuration.
- 🤔 Refactored TTS plugin parameters, separating any framework-coupled concepts (greatly reducing plugin encapsulation difficulty).
- 🤔 Refactored IAT plugin parameters, separating any framework-coupled concepts (greatly reducing plugin encapsulation difficulty).
- 🤔 Refactored LLM plugin parameters, separating any framework-coupled concepts (greatly reducing plugin encapsulation difficulty).
- 🤔 Extended the silence time when the user is not speaking.
- 💄 Improved error prompts for some configurations.
- 🐞 Fixed an issue where the first wake-up might prevent subsequent wake-ups.
- 🐞 Fixed potential confusion during conversations.
Client v1.0.0
- 🤔 Changed the parameter types of
esp_ai.onEvent()
toString
for easier usage (e.g., directly using==
to judge commands). - 💄 Improved voice wake-up accuracy.
2024-07-13
- 🆕 Added plugin development functionality, supporting custom plugins for the server.
- 🆕 Added client speaker sampling rate configuration.
- 🆕 Added Dolphin voice-over plugin
TTS
, supporting various voice tones. For details, see: https://www.ttson.cn/ - 💄 Improved audio stream splitting to reduce client speaker noise.
- 💄 Improved conversation fluency.
- 💄 Added a prompt sound before Xiao Ming listens.
2024-07-09
v1.0.0
- ✨ Refactored client code to header file format.
- 🆕 Added Volcano Engine
TTS
integration. - 🐞 Fixed an error when calling non-3.5 versions of
iFLYTEK LLM
. - 🐞 Fixed some server bugs.
2024-07-01
v0.0.1
- 🆕 Released the test version.