esp32~mp3播放实例解析

前言

esp32-ADF音频框架，Demo看了好几次，也只知道个大概。音频涉及的东西太多了，入门就先熟悉这个框架。
官方文档：https://docs.espressif.com/projects/esp-adf/en/latest/api-reference/index.html

Elements of the Audio Development Framework

框架中包含流、编解码器、音频处理相关的方法和概念。

Sample Organization of Elements in Audio Pipeline
将MP3解码器和I2S流两个元素添加进管道，解码器的输入是MP3文件数据流，I2S流将解码后的音频数据输出到片外，各应用程序之间通过事件接口通信。

1. 代码分析

static const char *TAG = "PLAY_MP3_FLASH";

/*
   To embed it in the app binary, the mp3 file is named
   in the component.mk COMPONENT_EMBED_TXTFILES variable.
*/
extern const uint8_t adf_music_mp3_start[] asm("_binary_adf_music_mp3_start");
extern const uint8_t adf_music_mp3_end[]   asm("_binary_adf_music_mp3_end");
static int adf_music_mp3_pos;

int mp3_music_read_cb(audio_element_handle_t el, char *buf, int len, TickType_t wait_time, void *ctx)
{
    int read_size = adf_music_mp3_end - adf_music_mp3_start - adf_music_mp3_pos;
    if (read_size == 0) {
        return AEL_IO_DONE;
    } else if (len < read_size) {
        read_size = len;
    }
    memcpy(buf, adf_music_mp3_start + adf_music_mp3_pos, read_size);
    adf_music_mp3_pos += read_size;
    return read_size;
}

void app_main(void)
{
    audio_pipeline_handle_t pipeline;   // 管道
    audio_element_handle_t i2s_stream_writer, mp3_decoder; // 音频元素

    esp_log_level_set("*", ESP_LOG_WARN);
    esp_log_level_set(TAG, ESP_LOG_INFO);

    ESP_LOGI(TAG, "[ 1 ] Start audio codec chip");
    audio_hal_codec_config_t audio_hal_codec_cfg =  AUDIO_HAL_ES8388_DEFAULT();
    audio_hal_handle_t hal = audio_hal_init(&audio_hal_codec_cfg, 0); // 初始化编解码驱动
    audio_hal_ctrl_codec(hal, AUDIO_HAL_CODEC_MODE_DECODE, AUDIO_HAL_CTRL_START); // 启动解码
    int player_volume;
    audio_hal_get_volume(hal, &player_volume);  // 获取音量

    ESP_LOGI(TAG, "[ 2 ] Create audio pipeline, add all elements to pipeline, and subscribe pipeline event");
    audio_pipeline_cfg_t pipeline_cfg = DEFAULT_AUDIO_PIPELINE_CONFIG();
    pipeline = audio_pipeline_init(&pipeline_cfg);  // 初始化管道
    mem_assert(pipeline);

    ESP_LOGI(TAG, "[2.1] Create mp3 decoder to decode mp3 file and set custom read callback");
    mp3_decoder_cfg_t mp3_cfg = DEFAULT_MP3_DECODER_CONFIG();
    mp3_decoder = mp3_decoder_init(&mp3_cfg);   // 初始化mp3 decoder元素
    audio_element_set_read_cb(mp3_decoder, mp3_music_read_cb, NULL);

    ESP_LOGI(TAG, "[2.2] Create i2s stream to write data to codec chip");
    i2s_stream_cfg_t i2s_cfg = I2S_STREAM_CFG_DEFAULT();
    i2s_cfg.type = AUDIO_STREAM_WRITER;
    i2s_stream_writer = i2s_stream_init(&i2s_cfg); // 初始化i2s stream元素

    ESP_LOGI(TAG, "[2.3] Register all elements to audio pipeline"); // 注册元素到管道中去
    audio_pipeline_register(pipeline, mp3_decoder, "mp3");
    audio_pipeline_register(pipeline, i2s_stream_writer, "i2s");

    ESP_LOGI(TAG, "[2.4] Link it together [mp3_music_read_cb]-->mp3_decoder-->i2s_stream-->[codec_chip]");
    audio_pipeline_link(pipeline, (const char *[]) {"mp3", "i2s"}, 2); // 将mp3_music_read_cb mp3_decoder i2s_stream codec_chip关联在一起

    ESP_LOGI(TAG, "[ 3 ] Initialize peripherals");
    esp_periph_config_t periph_cfg = { 0 };
    esp_periph_init(&periph_cfg);

    ESP_LOGI(TAG, "[3.1] Initialize Touch peripheral");
    periph_touch_cfg_t touch_cfg = {
        .touch_mask = TOUCH_SEL_SET | TOUCH_SEL_PLAY | TOUCH_SEL_VOLUP | TOUCH_SEL_VOLDWN,
        .tap_threshold_percent = 70,
    };
    esp_periph_handle_t touch_periph = periph_touch_init(&touch_cfg);   // 初始化按键

    ESP_LOGI(TAG, "[3.2] Start all peripherals");
    esp_periph_start(touch_periph);

    ESP_LOGI(TAG, "[ 4 ] Setup event listener");
    audio_event_iface_cfg_t evt_cfg = AUDIO_EVENT_IFACE_DEFAULT_CFG();
    audio_event_iface_handle_t evt = audio_event_iface_init(&evt_cfg);

    ESP_LOGI(TAG, "[4.1] Listening event from all elements of pipeline");
    audio_pipeline_set_listener(pipeline, evt);     // 设置按键事件

    ESP_LOGI(TAG, "[4.2] Listening event from peripherals");    // 设置外设监听事件
    audio_event_iface_set_listener(esp_periph_get_event_iface(), evt);

    ESP_LOGW(TAG, "[ 5 ] Tap touch buttons to control music player:");
    ESP_LOGW(TAG, "      [Play] to start, pause and resume, [Set] to stop.");
    ESP_LOGW(TAG, "      [Vol-] or [Vol+] to adjust volume.");

    while (1) {
        audio_event_iface_msg_t msg;
        esp_err_t ret = audio_event_iface_listen(evt, &msg, portMAX_DELAY);
        if (ret != ESP_OK) {
            ESP_LOGE(TAG, "[ * ] Event interface error : %d", ret);
            continue;
        }

        if (msg.source_type == AUDIO_ELEMENT_TYPE_ELEMENT && msg.source == (void *) mp3_decoder
            && msg.cmd == AEL_MSG_CMD_REPORT_MUSIC_INFO) {
            audio_element_info_t music_info = {0};
            audio_element_getinfo(mp3_decoder, &music_info);

            ESP_LOGI(TAG, "[ * ] Receive music info from mp3 decoder, sample_rates=%d, bits=%d, ch=%d",
                     music_info.sample_rates, music_info.bits, music_info.channels);

            audio_element_setinfo(i2s_stream_writer, &music_info);
            i2s_stream_set_clk(i2s_stream_writer, music_info.sample_rates, music_info.bits, music_info.channels);
            continue;
        }

        if (msg.source_type == PERIPH_ID_TOUCH
            && msg.cmd == PERIPH_TOUCH_TAP
            && msg.source == (void *)touch_periph) {

            if ((int) msg.data == TOUCH_PLAY) { // 改变播放状态
                ESP_LOGI(TAG, "[ * ] [Play] touch tap event");
                audio_element_state_t el_state = audio_element_get_state(i2s_stream_writer);
                switch (el_state) {
                    case AEL_STATE_INIT :
                        ESP_LOGI(TAG, "[ * ] Starting audio pipeline");
                        audio_pipeline_run(pipeline);
                        break;
                    case AEL_STATE_RUNNING :
                        ESP_LOGI(TAG, "[ * ] Pausing audio pipeline");
                        audio_pipeline_pause(pipeline);
                        break;
                    case AEL_STATE_PAUSED :
                        ESP_LOGI(TAG, "[ * ] Resuming audio pipeline");
                        audio_pipeline_resume(pipeline);
                        break;
                    case AEL_STATE_FINISHED :
                        ESP_LOGI(TAG, "[ * ] Rewinding audio pipeline");
                        audio_pipeline_stop(pipeline);
                        adf_music_mp3_pos = 0;
                        audio_pipeline_resume(pipeline);
                        break;
                    default :
                        ESP_LOGI(TAG, "[ * ] Not supported state %d", el_state);
                }
            } else if ((int) msg.data == TOUCH_SET) {
                ESP_LOGI(TAG, "[ * ] [Set] touch tap event");
                ESP_LOGI(TAG, "[ * ] Stopping audio pipeline");
                break;
            } else if ((int) msg.data == TOUCH_VOLUP) {
                ESP_LOGI(TAG, "[ * ] [Vol+] touch tap event");
                player_volume += 10;
                if (player_volume > 100) {
                    player_volume = 100;
                }
                audio_hal_set_volume(hal, player_volume);
                ESP_LOGI(TAG, "[ * ] Volume set to %d %%", player_volume);
            } else if ((int) msg.data == TOUCH_VOLDWN) {
                ESP_LOGI(TAG, "[ * ] [Vol-] touch tap event");
                player_volume -= 10;
                if (player_volume < 0) {
                    player_volume = 0;
                }
                audio_hal_set_volume(hal, player_volume);
                ESP_LOGI(TAG, "[ * ] Volume set to %d %%", player_volume);
            }
        }
    }

    ESP_LOGI(TAG, "[ 6 ] Stop audio_pipeline");
    audio_pipeline_terminate(pipeline);

    audio_pipeline_unregister(pipeline, mp3_decoder);
    audio_pipeline_unregister(pipeline, i2s_stream_writer);

    /* Terminate the pipeline before removing the listener */
    audio_pipeline_remove_listener(pipeline);

    /* Make sure audio_pipeline_remove_listener is called before destroying event_iface */
    audio_event_iface_destroy(evt);

    /* Release all resources */
    audio_pipeline_deinit(pipeline);
    audio_element_deinit(i2s_stream_writer);
    audio_element_deinit(mp3_decoder);
}

2. 思考

还是觉得音频太难了，这个框架中就涉及到很多东西。
参考下面作者学习：从play_mp3例程出发理解ESP32-ADF的使用方法

3. 实例中mp3文件是如何处理的？

/*
   To embed it in the app binary, the mp3 file is named
   in the component.mk COMPONENT_EMBED_TXTFILES variable.
*/
extern const uint8_t adf_music_mp3_start[] asm("_binary_adf_music_mp3_start");
extern const uint8_t adf_music_mp3_end[]   asm("_binary_adf_music_mp3_end");

在熟悉Demo的过程中对mp3文件的处理不是很清楚，查询网上资料也找不到好的答案。便直接提了个issue，挺幸运的，乐鑫的大佬有答复。根据下面的回复，mp3在实例中是编译转化为二进制的过程中就加入了文件的起始标志。这里又遇到内嵌汇编和Makefile了。