Implement audio

README.md

@@ -4,7 +4,7 @@
 
 This GIF / YouTube was created with this command: "editly [commonFeatures.json5](https://github.com/mifi/editly/blob/master/examples/commonFeatures.json5)". See [more examples here](https://github.com/mifi/editly/tree/master/examples#examples).
 
-**Editly** is a tool and framework for declarative NLE (**non-linear video editing**) using Node.js and ffmpeg. Editly allows you to easily and **programmatically create a video** from a **set of clips, images and titles**, with smooth transitions and music overlaid.
+**Editly** is a tool and framework for declarative NLE (**non-linear video editing**) using Node.js and ffmpeg. Editly allows you to easily and **programmatically create a video** from a **set of clips, images, audio and titles**, with smooth transitions and music overlaid.
 
 Editly has a simple CLI for quickly assembling a video from a set of clips or images, or you can use its more flexible JavaScript API.
 
@@ -143,7 +143,7 @@ Edit specs are JavaScript / JSON objects describing the whole edit operation wit
 | `width` | `--width` | Width which all media will be converted to | `640` | |
 | `height` | `--height` | Height which all media will be converted to | auto based on `width` and aspect ratio of **first video** | |
 | `fps` | `--fps` | FPS which all videos will be converted to | First video FPS or `25` | |
-| `audioFilePath` | `--audio-file-path` | Set an audio track to the whole output video | | |
+| `audioFilePath` | `--audio-file-path` | Set an audio track for the whole video | | |
 | `fast` | `--fast`, `-f` | Fast mode (low resolution and FPS, useful for getting a quick preview) | `false` | |
 | `defaults.layer.fontPath` | `--font-path` | Set default font to a .ttf | System font | |
 | `defaults.layer.*` | | Set any layer parameter that all layers will inherit | | |
@@ -152,7 +152,7 @@ Edit specs are JavaScript / JSON objects describing the whole edit operation wit
 | `defaults.transition.duration` | `--transition-duration` | Default transition duration | `0.5` | sec |
 | `defaults.transition.name` | `--transition-name` | Default transition type. See **Transition types** | `random` | |
 | `clips[]` | | List of clip objects that will be concatenated in sequence | | |
-| `clips[].duration` | | Clip duration. See `defaults.duration` | `defaults.duration` | |
+| `clips[].duration` | | Clip duration. See `defaults.duration`. If unset, the clip duration will be that of the first video layer. | `defaults.duration` | |
 | `clips[].transition` | | Specify transition at the **end** of this clip. See `defaults.transition` | `defaults.transition` | |
 | `clips[].layers[]` | | List of layers within the current clip that will be overlaid in their natural order (last layer on top) | | |
 | `clips[].layers[].type` | | Layer type, see below | | |
@@ -167,15 +167,27 @@ See [examples](https://github.com/mifi/editly/tree/master/examples) and [commonF
 
 #### Layer type 'video'
 
-For video layers, if parent `clip.duration` is specified, the video will be slowed/sped-up to match `clip.duration`. If `cutFrom`/`cutTo` is set, the resulting segment (`cutTo`-`cutFrom`) will be slowed/sped-up to fit `clip.duration`.
+For video layers, if parent `clip.duration` is specified, the video will be slowed/sped-up to match `clip.duration`. If `cutFrom`/`cutTo` is set, the resulting segment (`cutTo`-`cutFrom`) will be slowed/sped-up to fit `clip.duration`. If the layer has audio, it will be kept (and mixed with other audio layers if present.)
 
 | Parameter  | Description | Default | |
 |-|-|-|-|
 | `path` | Path to video file | | |
 | `resizeMode` | One of `cover`, `contain`, `stretch` | `contain` | |
 | `cutFrom` | Time value to cut from | `0` | sec |
-| `cutTo` | Time value to cut from | *end of video* | sec |
+| `cutTo` | Time value to cut to | *end of video* | sec |
 | `backgroundColor` | Background of letterboxing | `#000000` | |
+| `mixVolume` | Relative volume when mixing this video's audio track with others | `1` | |
+
+#### Layer type 'audio'
+
+Audio layers will be mixed together. If `cutFrom`/`cutTo` is set, the resulting segment (`cutTo`-`cutFrom`) will be slowed/sped-up to fit `clip.duration`. The slow down/speed-up operation is limited to `half speed` and `100x`.
+
+| Parameter  | Description | Default | |
+|-|-|-|-|
+| `path` | Path to audio file | | |
+| `cutFrom` | Time value to cut from | `0` | sec |
+| `cutTo` | Time value to cut to | `clip.duration` | sec |
+| `mixVolume` | Relative volume when mixing this audio track with others | `1` | |
 
 #### Layer type 'image'
 

audio.js

@@ -0,0 +1,135 @@
+const pMap = require('p-map');
+const { join, basename } = require('path');
+const execa = require('execa');
+const flatMap = require('lodash/flatMap');
+const fs = require('fs-extra');
+
+const { getFfmpegCommonArgs, getCutFromArgs, createConcatFile } = require('./ffmpeg');
+const { readFileStreams } = require('./util');
+
+module.exports = ({ ffmpegPath, ffprobePath, enableFfmpegLog, verbose }) => {
+  async function editAudio({ clips, tmpDir }) {
+    if (clips.length === 0) return undefined;
+
+    console.log('Extracting audio or creating silence from all clips');
+
+    const mergedAudioPath = join(tmpDir, 'audio-merged.flac');
+
+    const segments = await pMap(clips, async (clip, i) => {
+      const clipAudioPath = join(tmpDir, `clip${i}-audio.flac`);
+
+      const audioLayers = clip.layers.filter((layer) => ['audio', 'video'].includes(layer.type));
+
+      async function createSilence(outPath) {
+        if (verbose) console.log('create silence', clip.duration);
+        const args = [
+          '-f', 'lavfi', '-i', 'anullsrc=channel_layout=stereo:sample_rate=44100',
+          '-sample_fmt', 's32',
+          '-ar', '48000',
+          '-t', clip.duration,
+          '-c:a', 'flac',
+          '-y',
+          outPath,
+        ];
+        await execa(ffmpegPath, args);
+      }
+
+      if (audioLayers.length > 0) {
+        const processedAudioLayersRaw = await pMap(audioLayers, async (audioLayer, j) => {
+          const { path, cutFrom, audioCutTo, framePtsFactor } = audioLayer;
+
+          const streams = await readFileStreams(ffprobePath, path);
+          if (!streams.some((s) => s.codec_type === 'audio')) return undefined;
+
+          const layerAudioPath = join(tmpDir, `clip${i}-layer${j}-audio.flac`);
+
+          try {
+            let atempoFilter;
+            if (Math.abs(framePtsFactor - 1) > 0.01) {
+              if (verbose) console.log('audio framePtsFactor', framePtsFactor);
+              const atempo = (1 / framePtsFactor);
+              if (!(atempo >= 0.5 && atempo <= 100)) { // Required range by ffmpeg
+                console.warn(`Audio speed ${atempo} is outside accepted range, using silence (clip ${i})`);
+                return undefined;
+              }
+              atempoFilter = `atempo=${atempo}`;
+            }
+
+            const cutToArg = (audioCutTo - cutFrom) * framePtsFactor;
+
+            const args = [
+              ...getFfmpegCommonArgs({ enableFfmpegLog }),
+              ...getCutFromArgs({ cutFrom }),
+              '-i', path,
+              '-t', cutToArg,
+              '-sample_fmt', 's32',
+              '-ar', '48000',
+              '-map', 'a:0', '-c:a', 'flac',
+              ...(atempoFilter ? ['-filter:a', atempoFilter] : []),
+              '-y',
+              layerAudioPath,
+            ];
+
+            // console.log(args);
+            await execa(ffmpegPath, args);
+          } catch (err) {
+            if (verbose) console.error('Cannot extract audio from video', path, err);
+            // Fall back to silence
+            await createSilence(layerAudioPath);
+          }
+
+          return { layerAudioPath, audioLayer };
+        }, { concurrency: 4 });
+
+        const processedAudioLayers = processedAudioLayersRaw.filter((p) => p);
+
+        if (processedAudioLayers.length > 1) {
+          // Merge/mix all layer's audio
+
+          const weights = processedAudioLayers.map(({ audioLayer }) => (audioLayer.mixVolume != null ? audioLayer.mixVolume : 1));
+          const args = [
+            ...getFfmpegCommonArgs({ enableFfmpegLog }),
+            ...flatMap(processedAudioLayers, ({ layerAudioPath }) => ['-i', layerAudioPath]),
+            '-filter_complex', `amix=inputs=${processedAudioLayers.length}:duration=longest:weights=${weights.join(' ')}`,
+            '-c:a', 'flac',
+            '-y',
+            clipAudioPath,
+          ];
+
+          await execa(ffmpegPath, args);
+        } else if (processedAudioLayers.length > 0) {
+          await fs.rename(processedAudioLayers[0].layerAudioPath, clipAudioPath);
+        } else {
+          await createSilence(clipAudioPath);
+        }
+      } else {
+        await createSilence(clipAudioPath);
+      }
+
+      return basename(clipAudioPath);
+    }, { concurrency: 4 });
+
+    const concatFilePath = join(tmpDir, 'audio-segments.txt');
+
+    console.log('Combining audio', segments, concatFilePath);
+
+    await createConcatFile(segments, concatFilePath);
+
+    const args = [
+      ...getFfmpegCommonArgs({ enableFfmpegLog }),
+      '-f', 'concat', '-safe', '0',
+      '-i', concatFilePath,
+      '-c', 'flac',
+      '-y',
+      mergedAudioPath,
+    ];
+    await execa(ffmpegPath, args);
+
+    // TODO don't return audio if only silence?
+    return mergedAudioPath;
+  }
+
+  return {
+    editAudio,
+  };
+};

examples/alpha.json5

@@ -0,0 +1,14 @@
+{
+  // enableFfmpegLog: true,
+  outPath: './alpha.mp4',
+  clips: [
+    { duration: 2, layers: [
+      { type: 'video', path: './assets/lofoten.mp4', cutFrom: 0.4, cutTo: 2 },
+      { type: 'video', path: './assets/dancer1.webm', cutFrom: 0, cutTo: 6 },
+    ] },
+    { layers: [
+      { type: 'video', path: './assets/lofoten.mp4', cutFrom: 0.4, cutTo: 2 },
+      { type: 'video', path: './assets/dancer1.webm' },
+    ] },
+  ],
+}

examples/audio1.json5

@@ -0,0 +1,24 @@
+{
+  // enableFfmpegLog: true,
+  outPath: './audio1.mp4',
+  defaults: {
+    transition: null,
+  },
+  clips: [
+    { duration: 0.5, layers: [{ type: 'video', path: './assets/lofoten.mp4', cutFrom: 0.4, cutTo: 2 }] },
+
+    { layers: [
+      { type: 'title', text: 'test' },
+      { type: 'audio', path: './assets/High [NCS Release] - JPB  (No Copyright Music)-R8ZRCXy5vhA.m4a', cutFrom: 2, cutTo: 5 }] },
+
+    { layers: [
+      { type: 'video', path: './assets/lofoten.mp4', cutFrom: 12, cutTo: 14, mixVolume: 0 },
+      { type: 'audio', path: './assets/High [NCS Release] - JPB  (No Copyright Music)-R8ZRCXy5vhA.m4a', mixVolume: 0.1 }] },
+
+    { duration: 2, layers: [
+      { type: 'video', path: './assets/lofoten.mp4', cutFrom: 0.4, cutTo: 2 },
+      { type: 'audio', path: './assets/High [NCS Release] - JPB  (No Copyright Music)-R8ZRCXy5vhA.m4a', cutFrom: 2, cutTo: 3, mixVolume: 0.5 }] },
+
+    { duration: 1.8, layers: [{ type: 'video', path: './assets/lofoten.mp4', cutFrom: 10, cutTo: 11 }] },
+  ],
+}

examples/audio2.json5

@@ -0,0 +1,21 @@
+{
+  // enableFfmpegLog: true,
+  outPath: './audio2.mp4',
+  clips: [
+    { duration: 0.5, layers: [{ type: 'video', path: './assets/lofoten.mp4', cutFrom: 0.4, cutTo: 2 }] },
+
+    { layers: [
+      { type: 'title', text: 'test' },
+      { type: 'audio', path: './assets/High [NCS Release] - JPB  (No Copyright Music)-R8ZRCXy5vhA.m4a' }] },
+
+    { layers: [
+      { type: 'video', path: './assets/lofoten.mp4', cutFrom: 12, cutTo: 14, mixVolume: 0.7 },
+      { type: 'audio', path: './assets/High [NCS Release] - JPB  (No Copyright Music)-R8ZRCXy5vhA.m4a', mixVolume: 0.3 }] },
+
+    { layers: [
+      { type: 'video', path: './assets/lofoten.mp4', cutFrom: 0.4, cutTo: 2 },
+      { type: 'audio', path: './assets/High [NCS Release] - JPB  (No Copyright Music)-R8ZRCXy5vhA.m4a' }] },
+
+    { layers: [{ type: 'video', path: './assets/lofoten.mp4', cutFrom: 10, cutTo: 11 }] },
+  ],
+}

index.js

@@ -1,17 +1,18 @@
 const execa = require('execa');
 const assert = require('assert');
 const pMap = require('p-map');
-const { basename, join } = require('path');
+const { basename, join, dirname } = require('path');
 const flatMap = require('lodash/flatMap');
 const JSON5 = require('json5');
 const fs = require('fs-extra');
 
-const { parseFps, readFileInfo, multipleOf2 } = require('./util');
+const { parseFps, readVideoFileInfo, readAudioFileInfo, multipleOf2 } = require('./util');
 const { registerFont } = require('./sources/fabricFrameSource');
 const { createFrameSource } = require('./sources/frameSource');
 const { calcTransition } = require('./transitions');
 
 const GlTransitions = require('./glTransitions');
+const Audio = require('./audio');
 
 // Cache
 const loadedFonts = [];
@@ -42,7 +43,8 @@ module.exports = async (config = {}) => {
 
   const isGif = outPath.toLowerCase().endsWith('.gif');
 
-  const audioFilePath = isGif ? undefined : audioFilePathIn;
+  let audioFilePath;
+  if (!isGif) audioFilePath = audioFilePathIn;
 
   if (audioFilePath) await assertFileExists(audioFilePath);
 
@@ -126,66 +128,122 @@ module.exports = async (config = {}) => {
   }
 
   const clips = await pMap(clipsIn, async (clip, clipIndex) => {
-    const { transition: userTransition, duration: userDuration, layers } = clip;
+    const { transition: userTransition, duration: userClipDuration, layers } = clip;
 
     checkTransition(userTransition);
 
     const videoLayers = layers.filter((layer) => layer.type === 'video');
-    assert(videoLayers.length <= 1, 'Max 1 video per layer');
 
-    const userOrDefaultDuration = userDuration || defaults.duration;
-    if (videoLayers.length === 0) assert(userOrDefaultDuration, `Duration is required for clip ${clipIndex}`);
+    const userClipDurationOrDefault = userClipDuration || defaults.duration;
+    if (videoLayers.length === 0) assert(userClipDurationOrDefault, `Duration parameter is required for videoless clip ${clipIndex}`);
 
-    let duration = userOrDefaultDuration;
+    const transition = calcTransition(defaults, userTransition, clipIndex === clipsIn.length - 1);
 
-    const layersOut = flatMap(await pMap(layers, async (layerIn) => {
+    let layersOut = flatMap(await pMap(layers, async (layerIn) => {
       const layer = { ...defaults.layer, ...layerIn };
-      const { type } = layer;
+      const { type, path } = layer;
 
       if (type === 'video') {
-        const { cutFrom: cutFromIn, cutTo: cutToIn, path } = layer;
-        const fileInfo = await readFileInfo(ffprobePath, path);
-        const { duration: fileDuration, width: widthIn, height: heightIn, framerateStr, rotation } = fileInfo;
-        let cutFrom;
-        let cutTo;
-        let trimmedSourceDuration = fileDuration;
-        if (cutFromIn != null || cutToIn != null) {
-          cutFrom = Math.min(Math.max(0, cutFromIn || 0), fileDuration);
-          cutTo = Math.min(Math.max(cutFrom, cutToIn || fileDuration), fileDuration);
+        const { duration: fileDuration, width: widthIn, height: heightIn, framerateStr, rotation } = await readVideoFileInfo(ffprobePath, path);
+        let { cutFrom, cutTo } = layer;
+        if (!cutFrom) cutFrom = 0;
+        cutFrom = Math.max(cutFrom, 0);
+        cutFrom = Math.min(cutFrom, fileDuration);
+
+        if (!cutTo) cutTo = fileDuration;
+        cutTo = Math.max(cutTo, cutFrom);
+        cutTo = Math.min(cutTo, fileDuration);
         assert(cutFrom < cutTo, 'cutFrom must be lower than cutTo');
 
-          trimmedSourceDuration = cutTo - cutFrom;
-        }
-
-        // If user specified duration, means that should be the output duration
-        let framePtsFactor;
-        if (userDuration) {
-          duration = userDuration;
-          framePtsFactor = userDuration / trimmedSourceDuration;
-        } else {
-          duration = trimmedSourceDuration;
-          framePtsFactor = 1;
-        }
+        const inputDuration = cutTo - cutFrom;
 
         const isRotated = rotation === 90 || rotation === 270;
         const width = isRotated ? heightIn : widthIn;
         const height = isRotated ? widthIn : heightIn;
 
-        return { ...layer, cutFrom, cutTo, width, height, framerateStr, framePtsFactor };
+        // Compensate for transition duration
+        const audioCutTo = Math.max(cutFrom, cutTo - transition.duration);
+
+        return { ...layer, cutFrom, cutTo, audioCutTo, inputDuration, width, height, framerateStr };
       }
 
+      if (type === 'audio') return layer;
+
       return handleLayer(layer);
     }, { concurrency: 1 }));
 
-    const transition = calcTransition(defaults, userTransition);
+    let clipDuration = userClipDurationOrDefault;
+
+    const firstVideoLayer = layersOut.find((layer) => layer.type === 'video');
+    if (firstVideoLayer && !userClipDuration) clipDuration = firstVideoLayer.inputDuration;
+    assert(clipDuration);
+
+    layersOut = await pMap(layersOut, async (layer) => {
+      const { type, path } = layer;
+
+      if (type === 'audio') {
+        const { duration: fileDuration } = await readAudioFileInfo(ffprobePath, path);
+        let { cutFrom, cutTo } = layer;
+
+        console.log({ cutFrom, cutTo, fileDuration, clipDuration });
+
+        if (!cutFrom) cutFrom = 0;
+        cutFrom = Math.max(cutFrom, 0);
+        cutFrom = Math.min(cutFrom, fileDuration);
+
+        if (!cutTo) cutTo = cutFrom + clipDuration;
+        cutTo = Math.max(cutTo, cutFrom);
+        cutTo = Math.min(cutTo, fileDuration);
+        assert(cutFrom < cutTo, 'cutFrom must be lower than cutTo');
+
+        const inputDuration = cutTo - cutFrom;
+
+        const framePtsFactor = clipDuration / inputDuration;
+
+        // Compensate for transition duration
+        const audioCutTo = Math.max(cutFrom, cutTo - transition.duration);
+
+        return { ...layer, cutFrom, cutTo, audioCutTo, framePtsFactor };
+      }
+
+      if (layer.type === 'video') {
+        const { inputDuration } = layer;
+
+        let framePtsFactor;
+
+        // If user explicitly specified duration for clip, it means that should be the output duration of the video
+        if (userClipDuration) {
+          // Later we will speed up or slow down video using this factor
+          framePtsFactor = userClipDuration / inputDuration;
+        } else {
+          framePtsFactor = 1;
+        }
+
+        return { ...layer, framePtsFactor };
+      }
+
+      return layer;
+    });
 
     return {
       transition,
-      duration,
+      duration: clipDuration,
       layers: layersOut,
     };
   }, { concurrency: 1 });
 
+  const { editAudio } = Audio({ ffmpegPath, ffprobePath, enableFfmpegLog, verbose });
+
+  const outDir = dirname(outPath);
+  const tmpDir = join(outDir, 'editly-tmp');
+  if (verbose) console.log({ tmpDir });
+  await fs.remove(tmpDir);
+  await fs.mkdirp(tmpDir);
+
+  if (!audioFilePath) {
+    audioFilePath = await editAudio({ clips, tmpDir });
+  }
+
   if (verbose) console.log(JSON5.stringify(clips, null, 2));
 
   // Try to detect parameters from first video
@@ -340,7 +398,7 @@ module.exports = async (config = {}) => {
     const getTransitionFromClip = () => clips[transitionFromClipId];
     const getTransitionToClip = () => clips[getTransitionToClipId()];
 
-    const getSource = (clip, clipIndex) => createFrameSource({ clip, clipIndex, width, height, channels, verbose, ffmpegPath, enableFfmpegLog, framerateStr });
+    const getSource = (clip, clipIndex) => createFrameSource({ clip, clipIndex, width, height, channels, verbose, ffmpegPath, ffprobePath, enableFfmpegLog, framerateStr });
 
     const getTransitionToSource = async () => (getTransitionToClip() && getSource(getTransitionToClip(), getTransitionToClipId()));
     frameSource1 = await getSource(getTransitionFromClip(), transitionFromClipId);
@@ -436,6 +494,7 @@ module.exports = async (config = {}) => {
   } finally {
     if (frameSource1) await frameSource1.close();
     if (frameSource2) await frameSource2.close();
+    await fs.remove(tmpDir);
   }
 
   try {

sources/frameSource.js

@@ -5,11 +5,12 @@ const { rgbaToFabricImage, customFabricFrameSource, createCustomCanvasFrameSourc
 const createVideoFrameSource = require('./videoFrameSource');
 const { createGlFrameSource } = require('./glFrameSource');
 
-
-async function createFrameSource({ clip, clipIndex, width, height, channels, verbose, ffmpegPath, enableFfmpegLog, framerateStr }) {
+async function createFrameSource({ clip, clipIndex, width, height, channels, verbose, ffmpegPath, ffprobePath, enableFfmpegLog, framerateStr }) {
   const { layers, duration } = clip;
 
-  const layerFrameSources = await pMap(layers, async (layer, layerIndex) => {
+  const visualLayers = layers.filter((layer) => layer.type !== 'audio');
+
+  const layerFrameSources = await pMap(visualLayers, async (layer, layerIndex) => {
     const { type, ...params } = layer;
     console.log('createFrameSource', type, 'clip', clipIndex, 'layer', layerIndex);
 
@@ -29,15 +30,15 @@ async function createFrameSource({ clip, clipIndex, width, height, channels, ver
     const createFrameSourceFunc = frameSourceFuncs[type];
     assert(createFrameSourceFunc, `Invalid type ${type}`);
 
-    return createFrameSourceFunc({ ffmpegPath, width, height, duration, channels, verbose, enableFfmpegLog, framerateStr, params });
+    return createFrameSourceFunc({ ffmpegPath, ffprobePath, width, height, duration, channels, verbose, enableFfmpegLog, framerateStr, params });
   }, { concurrency: 1 });
 
   async function readNextFrame(progress) {
     const canvas = createFabricCanvas({ width, height });
 
     // eslint-disable-next-line no-restricted-syntax
-    for (const frameSource of layerFrameSources) {
-      const rgba = await frameSource.readNextFrame(progress, canvas);
+    for (const layerFrameSource of layerFrameSources) {
+      const rgba = await layerFrameSource.readNextFrame(progress, canvas);
       // Frame sources can either render to the provided canvas and return nothing
       // OR return an raw RGBA blob which will be drawn onto the canvas
       if (rgba) {

transitions.js

@@ -26,8 +26,8 @@ function getTransitionEasingFunction(easing, transitionName) {
   return (progress) => progress;
 }
 
-function calcTransition(defaults, transition) {
-  if (transition === null) return { duration: 0 };
+function calcTransition(defaults, transition, isLastClip) {
+  if (transition === null || isLastClip) return { duration: 0 };
 
   let transitionOrDefault = {
     name: (transition && transition.name) || (defaults.transition && defaults.transition.name),

util.js

@@ -1,4 +1,5 @@
 const execa = require('execa');
+const assert = require('assert');
 
 function parseFps(fps) {
   const match = typeof fps === 'string' && fps.match(/^([0-9]+)\/([0-9]+)$/);
@@ -10,17 +11,31 @@ function parseFps(fps) {
   return undefined;
 }
 
-async function readFileInfo(ffprobePath, p) {
+async function readDuration(ffprobePath, p) {
+  const { stdout } = await execa(ffprobePath, ['-v', 'error', '-show_entries', 'format=duration', '-of', 'default=noprint_wrappers=1:nokey=1', p]);
+  const parsed = parseFloat(stdout);
+  assert(!Number.isNaN(parsed));
+  return parsed;
+}
+
+async function readFileStreams(ffprobePath, p) {
   const { stdout } = await execa(ffprobePath, [
-    '-select_streams', 'v:0', '-show_entries', 'stream', '-of', 'json', p,
+    '-show_entries', 'stream', '-of', 'json', p,
   ]);
   const json = JSON.parse(stdout);
-  const stream = json.streams[0];
+  return json.streams;
+}
+
+async function readVideoFileInfo(ffprobePath, p) {
+  const streams = await readFileStreams(ffprobePath, p);
+  const stream = streams.find((s) => s.codec_type === 'video'); // TODO
+
+  const duration = await readDuration(ffprobePath, p);
 
   const rotation = stream.tags && stream.tags.rotate && parseInt(stream.tags.rotate, 10);
   return {
     // numFrames: parseInt(stream.nb_frames, 10),
-    duration: parseFloat(stream.duration, 10),
+    duration,
     width: stream.width, // TODO coded_width?
     height: stream.height,
     framerateStr: stream.r_frame_rate,
@@ -28,10 +43,30 @@ async function readFileInfo(ffprobePath, p) {
   };
 }
 
+async function readAudioFileInfo(ffprobePath, p) {
+  const duration = await readDuration(ffprobePath, p);
+
+  return { duration };
+}
+
+function toArrayInteger(buffer) {
+  if (buffer.length > 0) {
+    const data = new Uint8ClampedArray(buffer.length);
+    for (let i = 0; i < buffer.length; i += 1) {
+      data[i] = buffer[i];
+    }
+    return data;
+  }
+  return [];
+}
+
 const multipleOf2 = (x) => (x + (x % 2));
 
 module.exports = {
   parseFps,
-  readFileInfo,
+  readVideoFileInfo,
+  readAudioFileInfo,
   multipleOf2,
+  toArrayInteger,
+  readFileStreams,
 };