Prevent audio speeding up when trimming (#7599)

* fix trim region background colour and change showRedo -> show_redo

* tweak audio sampling logic

* add changeset

* add trim_region_color to py docs

* formatting

* use trim_region_color for trim border

---------

Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com>

.changeset/solid-grapes-reply.md

@@ -0,0 +1,6 @@
+---
+"@gradio/audio": patch
+"gradio": patch
+---
+
+fix:Prevent audio speeding up when trimming

gradio/components/audio.py

@@ -24,7 +24,8 @@ class WaveformOptions:
     A dataclass for specifying options for the waveform display in the Audio component. An instance of this class can be passed into the `waveform_options` parameter of `gr.Audio`.
     Parameters:
         waveform_color: The color (as a hex string or valid CSS color) of the full waveform representing the amplitude of the audio. Defaults to a light gray color.
-        waveform_progress_color: The color (as a hex string or valid CSS color) that the waveform fills with to as the audio plays. Defaults to an orange color.
+        waveform_progress_color: The color (as a hex string or valid CSS color) that the waveform fills with to as the audio plays. Defaults to the accent color.
+        trim_region_color: The color (as a hex string or valid CSS color) of the trim region. Defaults to the accent color.
         show_recording_waveform: Whether to show the waveform when recording audio. Defaults to True.
         show_controls: Whether to show the standard HTML audio player below the waveform when recording audio or playing recorded audio. Defaults to False.
         skip_length: The percentage (between 0 and 100) of the audio to skip when clicking on the skip forward / skip backward buttons. Defaults to 5.
@@ -33,6 +34,7 @@ class WaveformOptions:
 
     waveform_color: str | None = None
     waveform_progress_color: str | None = None
+    trim_region_color: str | None = None
     show_recording_waveform: bool = True
     show_controls: bool = False
     skip_length: int | float = 5
@@ -118,7 +120,7 @@ class Audio(
             editable: If True, allows users to manipulate the audio file if the component is interactive. Defaults to True.
             min_length: The minimum length of audio (in seconds) that the user can pass into the prediction function. If None, there is no minimum length.
             max_length: The maximum length of audio (in seconds) that the user can pass into the prediction function. If None, there is no maximum length.
-            waveform_options: A dictionary of options for the waveform display. Options include: waveform_color (str), waveform_progress_color (str), show_controls (bool), skip_length (int). Default is None, which uses the default values for these options.
+            waveform_options: A dictionary of options for the waveform display. Options include: waveform_color (str), waveform_progress_color (str), show_controls (bool), skip_length (int), trim_region_color (str). Default is None, which uses the default values for these options.
         """
         valid_sources: list[Literal["upload", "microphone"]] = ["upload", "microphone"]
         if sources is None:

js/audio/Index.svelte

@@ -114,6 +114,15 @@
 		resize: true
 	};
 
+	function set_trim_region_colour(): void {
+		document.documentElement.style.setProperty(
+			"--trim-region-color",
+			trim_region_settings.color || color_accent
+		);
+	}
+
+	set_trim_region_colour();
+
 	function handle_error({ detail }: CustomEvent<string>): void {
 		const [level, status] = detail.includes("Invalid file type")
 			? ["warning", "complete"]

js/audio/player/AudioPlayer.svelte

@@ -181,7 +181,7 @@
 				bind:mode
 				bind:trimDuration
 				bind:show_volume_slider
-				showRedo={interactive}
+				show_redo={interactive}
 				{handle_reset_value}
 				{waveform_options}
 				{trim_region_settings}

js/audio/recorder/AudioRecorder.svelte

@@ -242,7 +242,7 @@
 			{handle_trim_audio}
 			bind:trimDuration
 			bind:mode
-			showRedo
+			show_redo
 			{handle_reset_value}
 			{waveform_options}
 		/>

js/audio/shared/WaveformControls.svelte

@@ -14,7 +14,7 @@
 	export let audio_duration: number;
 	export let i18n: I18nFormatter;
 	export let playing: boolean;
-	export let showRedo = false;
+	export let show_redo = false;
 	export let interactive = false;
 	export let handle_trim_audio: (start: number, end: number) => void;
 	export let mode = "";
@@ -240,7 +240,7 @@
 
 	<div class="settings-wrapper">
 		{#if editable && interactive}
-			{#if showRedo && mode === ""}
+			{#if show_redo && mode === ""}
 				<button
 					class="action icon"
 					aria-label="Reset audio"

js/audio/shared/WaveformRecordControls.svelte

@@ -246,10 +246,23 @@
 	:global(::part(region)) {
 		border-radius: var(--radius-md);
 		height: 98% !important;
-		border: 1px solid var(--color-accent);
+		border: 1px solid var(--trim-region-color);
+		background-color: unset;
 		border-width: 1px 3px;
 	}
 
+	:global(::part(region))::after {
+		content: "";
+		position: absolute;
+		top: 0;
+		left: 0;
+		width: 100%;
+		height: 100%;
+		background: var(--trim-region-color);
+		opacity: 0.2;
+		border-radius: var(--radius-md);
+	}
+
 	:global(::part(region-handle)) {
 		width: 5px !important;
 		border: none;

js/audio/shared/audioBufferToWav.ts

@@ -44,11 +44,12 @@ export function audioBufferToWav(audioBuffer: AudioBuffer): Uint8Array {
 	offset += 4;
 
 	// Write PCM audio data
-	for (let i = 0; i < audioBuffer.numberOfChannels; i++) {
-		const channel = audioBuffer.getChannelData(i);
-		for (let j = 0; j < channel.length; j++) {
-			// Scaling Float32 to Int16
-			const sample = Math.max(-1, Math.min(1, channel[j]));
+	for (let i = 0; i < audioBuffer.length; i++) {
+		for (let channel = 0; channel < numOfChan; channel++) {
+			const sample = Math.max(
+				-1,
+				Math.min(1, audioBuffer.getChannelData(channel)[i])
+			);
 			view.setInt16(offset, sample * 0x7fff, true);
 			offset += 2;
 		}

test/test_components.py

@@ -844,6 +844,7 @@ class TestAudio:
                 "skip_length": 5,
                 "waveform_color": None,
                 "waveform_progress_color": None,
+                "trim_region_color": None,
             },
             "_selectable": False,
         }
@@ -897,6 +898,7 @@ class TestAudio:
                 "skip_length": 5,
                 "waveform_color": None,
                 "waveform_progress_color": None,
+                "trim_region_color": None,
             },
             "_selectable": False,
         }