speech_recognition/esp-sr
1
0
Fork 0
mirror of https://github.com/espressif/esp-sr.git synced 2025-09-15 15:28:44 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

doc/update README and CHANGELOG

Browse Source
This commit is contained in:
Sun Xiang Yu 2020-03-31 18:41:38 +08:00
parent f0853199fb
commit ad755e891b
4 changed files with 43 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
CHANGELOG.md
View File

@ -1,5 +1,11 @@
# Change log for esp-sr
## 0.6.0
update multinet_cn_1.4 and add CONTINUOUS RECOGNITION mode
improve hilexin wakeNet5X3 model(v5)
support IDFv4.0 build system
replace MAP algorithm with MASE(Mic Array Speech Enhancement) algorithm v1.0
## 0.5.0
add multinet1 English model v1.0
update multinet1 Chinese model v2.0

15
README.md
View File

@ -4,7 +4,7 @@ Espressif esp_sr provides basic algorithms for **Speech Recognition** applicatio
* The wake word detection model [WakeNet](wake_word_engine/README.md)
* The speech command recognition model [MultiNet](speech_command_recognition/README.md)
* Acoustic algorithm: AEC(Acoustic Echo Cancellation), VAD(Voice Activity Detection), AGC(Automatic Gain Control), NS(Noise Suppression)
* Acoustic algorithm: MASE(Mic Array Speech Enhancement), AEC(Acoustic Echo Cancellation), VAD(Voice Activity Detection), AGC(Automatic Gain Control), NS(Noise Suppression)
These algorithms are provided in the form of a component, so they can be integrated into your projects with minimum efforts.
@ -19,3 +19,16 @@ Currently, Espressif has not only provided an official wake word "Hi, Lexin" to
Espressif's speech command recognition model [MultiNet](speech_command_recognition/README.md) is specially designed to provide a flexible off-line speech command recognition model. With this model, you can easily add your own speech commands, eliminating the need to train model again.
Currently, Espressif **MultiNet** supports up to 100 Chinese or English speech commands, such as “打开空调” (Turn on the air conditioner) and “打开卧室灯” (Turn on the bedroom light).
## Acoustic algorithm
Espressif acoustic algorithm module is specially designed to improve speech recognition performance in far-field or noisy environments.
Currently, MASE algorithm supports 2-mic linear array and 3-mic circular array.
**In order to achieve optimal performance:**
* Please refer to hardware design [ESP32_Korvo](https://github.com/espressif/esp-skainet/tree/master/docs/zh_CN/hw-reference/esp32/user-guide-esp32-korvo-v1.1.md) or [ESP32-LyraT-Mini](https://docs.espressif.com/projects/esp-adf/en/latest/get-started/get-started-esp32-lyrat-mini.html).
* Please refer to software design [esp-skainet](https://github.com/espressif/esp-skainet).

13
README_cn.md
View File

@ -4,7 +4,7 @@ esp_sr 提供语音识别相关方向算法模型，目前主要包括三个模
* 唤醒词识别模型 [WakeNet](wake_word_engine/README_cn.md)
* 语音命令识别模型 [MultiNet](speech_command_recognition/README_cn.md)
* 声学算法：AEC(Acoustic Echo Cancellation), VAD(Voice Activity Detection), AGC(Automatic Gain Control), NS(Noise Suppression)
* 声学算法：MASE(Mic Array Speech Enhancement), AEC(Acoustic Echo Cancellation), VAD(Voice Activity Detection), AGC(Automatic Gain Control), NS(Noise Suppression)
这些算法以组件的形式提供，因此可以轻松地将它们集成到您的项目中。
@ -20,3 +20,14 @@ esp_sr 提供语音识别相关方向算法模型，目前主要包括三个模
目前模型支持类似“打开空调”，“打开卧室灯”等中文命令词识别和"Turn on/off the light" 等英文命令词识别，自定义语音命令词最大个数为 100。
## 声学算法
声学算法模块, 致力于提高复杂声学环境下的语音识别性能。MASE算法可有效改善远程或嘈杂环境下的语音识别性能。
目前MASE算法支持2-mic线性阵列和3-mic环形阵列。
**算法性能与硬件设计与软件配置息息相关，为达到最优性能:**
* 硬件设计建议参考 [ESP32_Korvo](https://github.com/espressif/esp-skainet/tree/master/docs/zh_CN/hw-reference/esp32/user-guide-esp32-korvo-v1.1.md) 或 [ESP32-LyraT-Mini](https://docs.espressif.com/projects/esp-adf/en/latest/get-started/get-started-esp32-lyrat-mini.html)。
* 软件设计建议参考 [esp-skainet](https://github.com/espressif/esp-skainet) 中相关示例。

3
acoustic_algorithm/include/esp_mase.h
View File

@ -67,7 +67,8 @@ mase_handle_t mase_create(int sample_rate, int frame_size, int array_type, float
*
* @return None
*
* @note Input is a multi-channel signal while the output is single-channel. For a 16-ms multi-channel input frame, the i-th point in the c-th channel should be indexed (i + c * 256).
* @note Input is a multi-channel signal while the output is single-channel.
* For a 16-ms multi-channel input frame, the i-th point in the c-th channel should be indexed (i + c * 256).
*
*/
void mase_process(mase_handle_t st, int16_t *in, int16_t *dsp_out);