@mega-yfue/eufy-sdk 0.5.0-beta.13 → 0.5.0-beta.14

package/dist/core/contracts.d.ts

@@ -551,16 +551,25 @@ export type AudioCodec = "aac-lc" | "aac-eld" | "g711a";
 /**
  * One audio access unit, carrying the codec the station declared for it.
  *
- * Sample rate and channel count are deliberately absent: they are not on the wire. The v6 app assumes
- * 16 kHz mono for every audio type rather than reading them, so the SDK does not invent fields the
- * device never sent.
+ * Sample rate and channel count are absent for `aac-lc` and `g711a`: they are not on the wire, and the v6 app
+ * assumes 16 kHz mono for every audio type rather than reading them. An `aac-lc` frame is one ADTS frame and
+ * `g711a` is raw A-law. An `aac-eld` frame is one raw access unit with no transport framing, so it carries the
+ * AudioSpecificConfig that decodes it.
  */
-export interface LiveAudioFrame {
+export type LiveAudioFrame = {
     /** Codec declared in the frame header. */
-    codec: AudioCodec;
-    /** Elementary-stream bytes (ADTS-framed for the two AAC profiles). */
+    codec: "aac-lc" | "g711a";
+    /** One ADTS frame for `aac-lc`, raw A-law samples for `g711a`. */
     data: Buffer;
-}
+    config?: undefined;
+} | {
+    /** Codec declared in the frame header. */
+    codec: "aac-eld";
+    /** One raw AAC-ELD access unit. */
+    data: Buffer;
+    /** AudioSpecificConfig (ISO/IEC 14496-3) that decodes `data`. */
+    config: Buffer;
+};
 /**
  * One fragmented-MP4 (CMAF) output unit from the native muxer. `init` (the `ftyp`+`moov` init
  * segment) is present exactly once, on the first fragment; every fragment carries a `moof`+`mdat`

package/dist/index.js

@@ -17931,6 +17931,7 @@ var AUDIO_HEADER_LEN = 16;
 var AUDIO_TYPE_OFFSET = 5;
 var SC4 = Buffer.from([0, 0, 0, 1]);
 var SC3 = Buffer.from([0, 0, 1]);
+var AAC_ELD_AUDIO_SPECIFIC_CONFIG = Buffer.from("f8f03000", "hex");
 var AUDIO_CODECS = { 0: "aac-lc", 2: "g711a", 7: "aac-eld" };
 var MAX_TRACED_DECODE_FAILURES = 3;
 var DEFAULT_KEEPALIVE_MS = 3e3;
@@ -18159,7 +18160,7 @@ var LiveStream = class extends EventEmitter3 {
           if (!codec) {
             this.logger.debug(`[live] dropping audio frame: unknown codec id ${codecId ?? "missing"}`);
           } else {
-            this.emit("audio", { codec, data: audio });
+            this.emit("audio", codec === "aac-eld" ? { codec, data: audio, config: AAC_ELD_AUDIO_SPECIFIC_CONFIG } : { codec, data: audio });
           }
         }
       }
@@ -20177,7 +20178,7 @@ var CODEC_BOXES = {
   h264: { sample: "avc1", config: "avcC" },
   h265: { sample: "hvc1", config: "hvcC" }
 };
-var AAC_ELD_SAMPLES_PER_FRAME = 512;
+var AAC_ELD_SAMPLES_PER_FRAME = 480;
 var ADTS_FREQUENCY_INDEX_16K = 8;
 var ADTS_CHANNELS_MONO = 1;
 var Fmp4Muxer = class {
@@ -20195,6 +20196,8 @@ var Fmp4Muxer = class {
   lastVideoTimestampMs;
   audioRequested;
   audioCodec;
+  /** Decoder config of an AAC-ELD track, taken from its frames; absent for AAC-LC, whose `esds` config is fixed. */
+  audioConfig;
   audioDisabled = false;
   audioSamples = [];
   audioBaseDecodeTime = 0;
@@ -20261,7 +20264,10 @@ var Fmp4Muxer = class {
     }
     return void 0;
   }
-  /** Add one station-declared AAC access unit to the audio track. G.711 remains available via `live()`. */
+  /**
+   * Add one station-declared AAC access unit to the audio track: an ADTS frame for AAC-LC, a raw access unit
+   * described by its frame's config for AAC-ELD. G.711 remains available via `live()`.
+   */
   pushAudio(frame, timestampMs = now()) {
     if (!this.audioRequested || this.audioDisabled)
       return void 0;
@@ -20277,12 +20283,19 @@ var Fmp4Muxer = class {
     if (this.audioCodec && this.audioCodec !== frame.codec) {
       return this.disableAudio();
     }
-    const header = parseAdtsHeader(frame.data);
-    if (!header || header.frameLength > frame.data.length) {
-      throw new Error(`fMP4 muxer: ${frame.codec} frame is not a complete ADTS access unit`);
-    }
-    if (header.frequencyIndex !== ADTS_FREQUENCY_INDEX_16K || header.channels !== ADTS_CHANNELS_MONO) {
-      return this.disableAudio();
+    let sample;
+    if (frame.codec === "aac-eld") {
+      this.audioConfig = frame.config;
+      sample = frame.data;
+    } else {
+      const header = parseAdtsHeader(frame.data);
+      if (!header || header.frameLength > frame.data.length) {
+        throw new Error(`fMP4 muxer: ${frame.codec} frame is not a complete ADTS access unit`);
+      }
+      if (header.frequencyIndex !== ADTS_FREQUENCY_INDEX_16K || header.channels !== ADTS_CHANNELS_MONO) {
+        return this.disableAudio();
+      }
+      sample = frame.data.subarray(header.headerLength, header.frameLength);
     }
     this.audioCodec = frame.codec;
     this.firstAudioTimestampMs ??= timestampMs;
@@ -20294,7 +20307,7 @@ var Fmp4Muxer = class {
       this.audioSamples[this.audioSamples.length - 1].duration = measured;
     }
     this.audioSamples.push({
-      data: frame.data.subarray(header.headerLength, header.frameLength),
+      data: sample,
       duration: frame.codec === "aac-eld" ? AAC_ELD_SAMPLES_PER_FRAME : AAC_SAMPLES_PER_FRAME,
       keyframe: true
     });
@@ -20567,7 +20580,7 @@ var Fmp4Muxer = class {
     return box("mp4a", head, this.esds());
   }
   esds() {
-    const config = this.audioCodec === "aac-eld" ? Buffer.from([248, 240, 32]) : Buffer.from([20, 8]);
+    const config = this.audioConfig ?? Buffer.from([20, 8]);
     const specific = descriptor(5, config);
     const decoder = descriptor(4, Buffer.concat([Buffer.from([64, 21, 0, 0, 0]), u32(64e3), u32(32e3), specific]));
     const sl = descriptor(6, Buffer.from([2]));