Merge pull request #11 from Eyevinn/segmented-audio-encode

Segmented audio encode

Author: grusell

checks.gradle

@@ -19,6 +19,7 @@ jacocoTestCoverageVerification {
                     '*FfmpegExecutor.runFfmpeg$lambda$?(java.lang.Process)',
                     '*FfmpegExecutorKt.getProgressRegex()',
                     '*FilterSettings.*',
+                    'se.svt.oss.encore.service.EncoreService.handleProgress.1.1.emit(int, kotlin.coroutines.Continuation)',
             ]
             limit {
                 counter = 'LINE'

encore-common/src/main/kotlin/se/svt/oss/encore/config/EncodingProperties.kt

@@ -6,10 +6,11 @@ package se.svt.oss.encore.config
 
 import org.springframework.boot.context.properties.NestedConfigurationProperty
 import org.springframework.core.io.Resource
+import se.svt.oss.encore.model.AudioEncodingMode
 import se.svt.oss.encore.model.profile.ChannelLayout
 
 data class SegmentedEncodingProperties(
-    val enabledForAudio: Boolean = true,
+    val audioEncodingMode: AudioEncodingMode = AudioEncodingMode.ENCODE_WITH_VIDEO,
 )
 
 data class EncodingProperties(

encore-common/src/main/kotlin/se/svt/oss/encore/model/AudioEncodingMode.kt

@@ -0,0 +1,28 @@
+// SPDX-FileCopyrightText: 2025 Eyevinn Technology AB
+//
+// SPDX-License-Identifier: EUPL-1.2
+
+package se.svt.oss.encore.model
+
+/**
+ * Defines how audio should be encoded when using segmented encoding.
+ */
+enum class AudioEncodingMode {
+    /**
+     * Encode audio and video together in the same segments.
+     * Creates N tasks of type AUDIOVIDEOSEGMENT.
+     */
+    ENCODE_WITH_VIDEO,
+
+    /**
+     * Encode audio separately from video as a single full-length file (not segmented).
+     * Creates 1 AUDIOFULL task + N VIDEOSEGMENT tasks.
+     */
+    ENCODE_SEPARATELY_FULL,
+
+    /**
+     * Encode audio separately from video, with both audio and video segmented.
+     * Creates N AUDIOSEGMENT tasks + N VIDEOSEGMENT tasks (2N total tasks).
+     */
+    ENCODE_SEPARATELY_SEGMENTED,
+}

encore-common/src/main/kotlin/se/svt/oss/encore/model/EncoreJob.kt

@@ -30,24 +30,44 @@ data class SegmentedEncodingInfo(
     )
     val segmentLength: Double,
     @field:Schema(
-        description = "Number of segments",
+        description = "Number of video segments",
         nullable = false,
         readOnly = true,
     )
     val numSegments: Int,
     @field:Schema(
-        description = "Number of encoding tasks used for this job. This is either the number of segments, or the number of segments + 1 if separate audio encode is used.",
+        description = "Number of encoding tasks used for this job. This will be equal to numSegments plus numAudioSegments",
         nullable = false,
         readOnly = true,
     )
     val numTasks: Int,
     @field:Schema(
-        description = "Indicates if audio is encoded in segments. Otherwise a separate task will be used to encode the full audio.",
-        example = "true",
+        description = "The audio encoding mode used for this job.",
+        example = "ENCODE_WITH_VIDEO",
         nullable = false,
         readOnly = true,
     )
-    val segmentedAudioEncode: Boolean,
+    val audioEncodingMode: AudioEncodingMode,
+    @field:Schema(
+        description = "Audio segment padding in seconds (added at start/end of segments to avoid artifacts). Only relevant in ENCODE_SEPARATELY_SEGMENTED mode.",
+        example = "0.04267",
+        nullable = false,
+        readOnly = true,
+    )
+    val audioSegmentPadding: Double = 0.0,
+    @field:Schema(
+        description = "Length of each audio segment in seconds. Only relevant in ENCODE_SEPARATELY_SEGMENTED mode.",
+        example = "256.0",
+        nullable = false,
+        readOnly = true,
+    )
+    val audioSegmentLength: Double = 0.0,
+    @field:Schema(
+        description = "Number of audio segments",
+        nullable = false,
+        readOnly = true,
+    )
+    val numAudioSegments: Int,
 )
 
 @Validated
@@ -137,12 +157,20 @@ data class EncoreJob(
     val segmentLength: Double? = null,
 
     @field:Schema(
-        description = "If true, and segmented encoding i used, audio will be encoded in segments. Otherwise a separate task will be used to encode the full audio.",
-        example = "true",
-        defaultValue = "true",
+        description = "Defines how audio should be encoded when using segmented encoding. ENCODE_WITH_VIDEO: audio and video together in segments; ENCODE_SEPARATELY_FULL: audio separately as full file; ENCODE_SEPARATELY_SEGMENTED: audio separately in segments.",
+        example = "ENCODE_WITH_VIDEO",
+        defaultValue = "ENCODE_WITH_VIDEO",
+        nullable = true,
+    )
+    val audioEncodingMode: AudioEncodingMode? = null,
+
+    @field:Schema(
+        description = "Length of audio segments in seconds when using ENCODE_SEPARATELY_SEGMENTED mode. If not specified, a value close to 256s will be calculated that is a multiple of the audio frame size.",
+        example = "256.0",
         nullable = true,
     )
-    val segmentedEncodingEnabledForAudio: Boolean? = null,
+    @field:Positive
+    val audioSegmentLength: Double? = null,
 
     @field:Schema(
         description = "Properties for segmented encoding, or null if not used",

encore-common/src/main/kotlin/se/svt/oss/encore/model/profile/AudioEncode.kt

@@ -17,7 +17,7 @@ import se.svt.oss.encore.model.output.AudioStreamEncode
 import se.svt.oss.encore.model.output.Output
 
 data class AudioEncode(
-    val codec: String = "libfdk_aac",
+    override val codec: String = "libfdk_aac",
     val bitrate: String? = null,
     val samplerate: Int = 48000,
     val channelLayout: ChannelLayout = ChannelLayout.CH_LAYOUT_STEREO,

encore-common/src/main/kotlin/se/svt/oss/encore/model/profile/AudioEncoder.kt

@@ -12,7 +12,8 @@ private val log = KotlinLogging.logger { }
 abstract class AudioEncoder : OutputProducer {
 
     abstract val optional: Boolean
-    abstract val enabled: Boolean
+    abstract val codec: String
+    abstract override val enabled: Boolean
 
     fun logOrThrow(message: String): Output? {
         if (optional || !enabled) {

encore-common/src/main/kotlin/se/svt/oss/encore/model/profile/OutputProducer.kt

@@ -21,5 +21,6 @@ import se.svt.oss.encore.model.output.Output
     JsonSubTypes.Type(value = ThumbnailMapEncode::class, name = "ThumbnailMapEncode"),
 )
 interface OutputProducer {
+    val enabled: Boolean
     fun getOutput(job: EncoreJob, encodingProperties: EncodingProperties, filterSettings: FilterSettings): Output?
 }

encore-common/src/main/kotlin/se/svt/oss/encore/model/profile/SimpleAudioEncode.kt

@@ -13,7 +13,7 @@ import se.svt.oss.encore.model.output.AudioStreamEncode
 import se.svt.oss.encore.model.output.Output
 
 data class SimpleAudioEncode(
-    val codec: String = "libfdk_aac",
+    override val codec: String = "libfdk_aac",
     val bitrate: String? = null,
     val samplerate: Int? = null,
     val suffix: String = "_$codec",

encore-common/src/main/kotlin/se/svt/oss/encore/model/profile/ThumbnailEncode.kt

@@ -25,7 +25,7 @@ data class ThumbnailEncode(
     val suffixZeroPad: Int = 2,
     val inputLabel: String = DEFAULT_VIDEO_LABEL,
     val optional: Boolean = false,
-    val enabled: Boolean = true,
+    override val enabled: Boolean = true,
     val intervalSeconds: Double? = null,
     val decodeOutput: Int? = null,
 ) : OutputProducer {

encore-common/src/main/kotlin/se/svt/oss/encore/model/profile/ThumbnailMapEncode.kt

@@ -26,7 +26,7 @@ data class ThumbnailMapEncode(
     val rows: Int = 20,
     val quality: Int = 5,
     val optional: Boolean = true,
-    val enabled: Boolean = true,
+    override val enabled: Boolean = true,
     val suffix: String = "_${cols}x${rows}_${tileWidth}x${tileHeight}_thumbnail_map",
     val format: String = "jpg",
     val inputLabel: String = DEFAULT_VIDEO_LABEL,