Lucas RayanandClaude Sonnet 5.5
5a9fe676fc
CI / Lint e auditorias (push) Successful in 1m10s
CI / Trivy (dependências e segredos) (push) Successful in 12s
CI / Imagem Docker (linux/arm64) (push) Successful in 2m39s
Release / Plano (reaproveitar a imagem?) (push) Successful in 6s
Release / ci (push) Skipped
Release / Publica a imagem (linux/arm64) (push) Successful in 18s
Release / Cria a release no Gitea (push) Successful in 5s
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
99 lines
3.2 KiB
C
99 lines
3.2 KiB
C
/*
|
||
* 面向 ESP32-S3 / ESP32-P4 的 ESP-SR 示例
|
||
* 离线本地语音识别（WakeNet + MultiNet）
|
||
*
|
||
* 依赖（idf_component.yml）：
|
||
* espressif/esp-sr: "^1.5.0"
|
||
*/
|
||
|
||
#include <stdio.h>
|
||
#include <stdlib.h>
|
||
#include "freertos/FreeRTOS.h"
|
||
#include "freertos/task.h"
|
||
#include "esp_log.h"
|
||
#include "esp_wn_iface.h"
|
||
#include "esp_wn_models.h"
|
||
#include "esp_mn_iface.h"
|
||
#include "esp_mn_models.h"
|
||
#include "esp_afe_sr_iface.h"
|
||
#include "esp_afe_sr_models.h"
|
||
|
||
static const char *TAG = "app_voice_sr";
|
||
|
||
void feed_audio_task(void *arg)
|
||
{
|
||
esp_afe_sr_data_t *afe_data = (esp_afe_sr_data_t *)arg;
|
||
int audio_chunksize = afe_handle->get_feed_chunksize(afe_data);
|
||
int16_t *feed_buff = (int16_t *)malloc(audio_chunksize * sizeof(int16_t) * 2);
|
||
|
||
while (1) {
|
||
// 从开发板麦克风读取 I2S 音频
|
||
// i2s_read(I2S_NUM_0, feed_buff, ...);
|
||
|
||
afe_handle->feed(afe_data, feed_buff);
|
||
vTaskDelay(pdMS_TO_TICKS(10));
|
||
}
|
||
free(feed_buff);
|
||
vTaskDelete(NULL);
|
||
}
|
||
|
||
void speech_recognition_task(void *arg)
|
||
{
|
||
esp_afe_sr_data_t *afe_data = (esp_afe_sr_data_t *)arg;
|
||
const esp_afe_sr_iface_t *afe_handle = &ESP_AFE_SR_HANDLE;
|
||
|
||
// 初始化语音命令模型（MultiNet）
|
||
const esp_mn_iface_t *multinet = &ESP_MN_HANDLE;
|
||
model_iface_data_t *model_data = multinet->create(&MULTINET_MODEL, 6000);
|
||
|
||
ESP_LOGI(TAG, "Waiting for the wake word (e.g. 'Hi, ESP')...");
|
||
|
||
while (1) {
|
||
afe_fetch_result_t *res = afe_handle->fetch(afe_data);
|
||
if (!res || res->ret_value == ESP_FAIL) {
|
||
continue;
|
||
}
|
||
|
||
// 1. 唤醒词（Wake Word）检测
|
||
if (res->wake_word_detected == WAKENET_DETECTED) {
|
||
ESP_LOGW(TAG, ">>> WAKE WORD DETECTED! Listening channel open.");
|
||
}
|
||
|
||
// 2. 本地命令识别
|
||
if (res->raw_data_size > 0) {
|
||
esp_mn_state_t mn_state = multinet->detect(model_data, res->data);
|
||
|
||
if (mn_state == ESP_MN_STATE_DETECTED) {
|
||
esp_mn_results_t *mn_res = multinet->get_results(model_data);
|
||
ESP_LOGI(TAG, ">>> COMMAND RECOGNIZED: ID #%d (matching phrase)", mn_res->command_id[0]);
|
||
|
||
// 操作示例
|
||
if (mn_res->command_id[0] == 1) {
|
||
ESP_LOGI(TAG, "Action: turn on the living room light");
|
||
} else if (mn_res->command_id[0] == 2) {
|
||
ESP_LOGI(TAG, "Action: turn off the living room light");
|
||
}
|
||
} else if (mn_state == ESP_MN_STATE_TIMEOUT) {
|
||
ESP_LOGI(TAG, "Timeout. Returning to Wake Word listening mode.");
|
||
}
|
||
}
|
||
}
|
||
vTaskDelete(NULL);
|
||
}
|
||
|
||
void app_main(void)
|
||
{
|
||
ESP_LOGI(TAG, "Starting ESP-SR Speech Recognition Engine");
|
||
|
||
// 配置音频前端流水线（AEC、BSS、VAD、WakeNet）
|
||
afe_config_t afe_config = AFE_CONFIG_DEFAULT();
|
||
afe_config.wakenet_init = true;
|
||
afe_config.voice_communication_init = false;
|
||
|
||
esp_afe_sr_data_t *afe_data = ESP_AFE_SR_HANDLE.create_from_config(&afe_config);
|
||
|
||
// 创建音频输入和识别任务
|
||
xTaskCreatePinnedToCore(feed_audio_task, "feed_task", 4096, afe_data, 5, NULL, 0);
|
||
xTaskCreatePinnedToCore(speech_recognition_task, "sr_task", 8192, afe_data, 5, NULL, 1);
|
||
}
|