Skip to content

Speakr is a personal, self-hosted web application designed for transcribing audio recordings

Python

Server-render share page transcript for accessibility

Murtaza Nasirf0b8117417f2
diff --git a/src/api/shares.py b/src/api/shares.py
index b839920e..e0ae6dac 100644
--- a/src/api/shares.py
+++ b/src/api/shares.py
@@ -9,6 +9,7 @@ This blueprint handles:
import os
import re
+import json
from flask import Blueprint, render_template, request, redirect, url_for, jsonify, send_file, current_app
from flask_login import login_required, current_user
@@ -34,6 +35,85 @@ def init_shares_helpers(_has_recording_access):
has_recording_access = _has_recording_access
+def process_transcription_for_template(transcription_str):
+ """
+ Process transcription JSON into a format ready for server-side rendering.
+
+ Returns a dict with:
+ - is_json: bool - whether transcription is valid JSON
+ - has_speakers: bool - whether diarization data exists
+ - segments: list - processed segments with speaker info and colors
+ - speakers: list - unique speakers with colors
+ - plain_text: str - plain text version for non-JSON or fallback
+ """
+ if not transcription_str:
+ return {'is_json': False, 'has_speakers': False, 'segments': [], 'speakers': [], 'plain_text': ''}
+
+ try:
+ data = json.loads(transcription_str)
+ except (json.JSONDecodeError, TypeError):
+ # Plain text transcription
+ return {
+ 'is_json': False,
+ 'has_speakers': False,
+ 'segments': [],
+ 'speakers': [],
+ 'plain_text': transcription_str
+ }
+
+ if not isinstance(data, list):
+ return {
+ 'is_json': False,
+ 'has_speakers': False,
+ 'segments': [],
+ 'speakers': [],
+ 'plain_text': transcription_str
+ }
+
+ # Check if diarized (has speaker info)
+ has_speakers = any(seg.get('speaker') for seg in data)
+
+ # Get unique speakers and assign colors
+ speakers = []
+ speaker_colors = {}
+ if has_speakers:
+ unique_speakers = list(dict.fromkeys(seg.get('speaker') for seg in data if seg.get('speaker')))
+ for i, speaker in enumerate(unique_speakers):
+ color = f'speaker-color-{(i % 8) + 1}'
+ speaker_colors[speaker] = color
+ speakers.append({'name': speaker, 'color': color})
+
+ # Process segments
+ segments = []
+ last_speaker = None
+ for seg in data:
+ speaker = seg.get('speaker', '')
+ segment = {
+ 'text': seg.get('sentence', ''),
+ 'speaker': speaker,
+ 'start_time': seg.get('start_time') or seg.get('startTime', ''),
+ 'end_time': seg.get('end_time') or seg.get('endTime', ''),
+ 'color': speaker_colors.get(speaker, 'speaker-color-1'),
+ 'show_speaker': speaker != last_speaker
+ }
+ segments.append(segment)
+ last_speaker = speaker
+
+ # Build plain text version
+ if has_speakers:
+ plain_text = '\n'.join(f"[{seg['speaker']}]: {seg['text']}" for seg in segments)
+ else:
+ plain_text = '\n'.join(seg['text'] for seg in segments)
+
+ return {
+ 'is_json': True,
+ 'has_speakers': has_speakers,
+ 'segments': segments,
+ 'speakers': speakers,
+ 'plain_text': plain_text
+ }
+
+
# --- Public Sharing Routes ---
@shares_bp.route('/share/<string:public_id>', methods=['GET'])
@@ -42,6 +122,9 @@ def view_shared_recording(public_id):
share = Share.query.filter_by(public_id=public_id).first_or_404()
recording = share.recording
+ # Process transcription for server-side rendering
+ processed_transcript = process_transcription_for_template(recording.transcription)
+
# Create a limited dictionary for the public view
recording_data = {
'id': recording.id,
@@ -59,7 +142,7 @@ def view_shared_recording(public_id):
'audio_duration': recording.get_audio_duration()
}
- return render_template('share.html', recording=recording_data)
+ return render_template('share.html', recording=recording_data, transcript=processed_transcript)
@shares_bp.route('/share/audio/<string:public_id>')
diff --git a/templates/share.html b/templates/share.html
index cec0105e..64cef94f 100644
--- a/templates/share.html
+++ b/templates/share.html
@@ -216,7 +216,7 @@
<div v-show="activeTab === 'transcription'" class="w-full flex-1 flex flex-col overflow-hidden">
<!-- Transcription Controls -->
<div class="flex items-center justify-between mb-4 flex-shrink-0">
- <div v-if="processedTranscription.hasDialogue" class="view-mode-toggle">
+ <div v-if="hasDialogue" class="view-mode-toggle">
<button @click="transcriptView = 'simple'"
:class="['toggle-button', transcriptView === 'simple' ? 'active' : '']">
<i class="fas fa-list mr-1"></i>Simple
@@ -232,65 +232,64 @@
</div>
<!-- Speaker Legend (only for bubble view) -->
- <div v-if="processedTranscription.hasDialogue && processedTranscription.speakers.length > 0 && transcriptView === 'bubble'"
+ {% if transcript.has_speakers and transcript.speakers %}
+ <div v-show="transcriptView === 'bubble'"
:class="['speaker-legend', legendExpanded ? 'expanded' : '', 'flex-shrink-0', 'mb-4']">
<div class="speaker-legend-header" @click="legendExpanded = !legendExpanded">
<div class="speaker-legend-title">
<i class="fas fa-users"></i>
Speakers
- <span class="speaker-count-indicator">(${processedTranscription.speakers.length})</span>
+ <span class="speaker-count-indicator">({{ transcript.speakers|length }})</span>
</div>
<i class="fas fa-chevron-down speaker-legend-toggle"></i>
</div>
<div class="speaker-legend-content">
- <div v-for="(speaker, index) in processedTranscription.speakers"
- :key="index"
- :class="['speaker-legend-item', speaker.color]">
- <span class="speaker-name">${speaker.name}</span>
+ {% for speaker in transcript.speakers %}
+ <div class="speaker-legend-item {{ speaker.color }}">
+ <span class="speaker-name">{{ speaker.name }}</span>
</div>
+ {% endfor %}
</div>
</div>
+ {% endif %}
- <!-- Transcription Content - Scrollable Box -->
+ <!-- Transcription Content - Scrollable Box (Server-rendered for accessibility) -->
<div class="w-full flex-1 overflow-y-auto transcription-box" @click="handleTranscriptClick">
- <!-- Simple View -->
- <div v-if="transcriptView === 'simple'" class="transcription-simple-view">
- <div v-for="segment in processedTranscription.simpleSegments"
- :key="segment.startTime || Math.random()"
- class="speaker-segment cursor-pointer hover:bg-[var(--bg-accent-hover)] p-2 rounded transition-colors mb-2"
- :data-start-time="segment.startTime"
- @click="seekAudio(segment.startTime)"
+ {% if transcript.is_json and transcript.segments %}
+ <!-- Simple View (server-rendered) -->
+ <div v-show="transcriptView === 'simple'" class="transcription-simple-view">
+ {% for segment in transcript.segments %}
+ <div class="speaker-segment cursor-pointer hover:bg-[var(--bg-accent-hover)] p-2 rounded transition-colors mb-2"
+ data-start-time="{{ segment.start_time }}"
style="margin-bottom: 0.5rem !important;">
- <div v-if="segment.showSpeaker" :class="['speaker-tablet', segment.color]">
- ${segment.speaker}
- </div>
- <div class="speaker-text">
- ${segment.sentence}
- </div>
+ {% if segment.show_speaker and segment.speaker %}
+ <div class="speaker-tablet {{ segment.color }}">{{ segment.speaker }}</div>
+ {% endif %}
+ <div class="speaker-text">{{ segment.text }}</div>
</div>
+ {% endfor %}
</div>
- <!-- Bubble View -->
- <div v-else-if="transcriptView === 'bubble'" class="transcription-with-speakers">
- <div v-for="(row, rowIndex) in processedTranscription.bubbleRows"
- :key="rowIndex"
- :class="['bubble-row', row.isMe ? 'speaker-me' : '']">
- <div v-for="bubble in row.bubbles"
- :key="bubble.startTime || Math.random()"
- :class="['speaker-bubble', bubble.color, row.isMe ? 'speaker-me' : '', 'cursor-pointer']"
- :data-start-time="bubble.startTime"
- @click="seekAudio(bubble.startTime)">
- <div class="speaker-bubble-content">
- ${bubble.sentence}
- </div>
+ <!-- Bubble View (server-rendered) -->
+ <div v-show="transcriptView === 'bubble'" class="transcription-with-speakers">
+ {% set ns = namespace(last_speaker=None) %}
+ {% for segment in transcript.segments %}
+ {% if segment.speaker != ns.last_speaker %}
+ {% if not loop.first %}</div>{% endif %}
+ <div class="bubble-row {% if segment.speaker and 'me' in segment.speaker|lower %}speaker-me{% endif %}">
+ {% endif %}
+ <div class="speaker-bubble {{ segment.color }} {% if segment.speaker and 'me' in segment.speaker|lower %}speaker-me{% endif %} cursor-pointer"
+ data-start-time="{{ segment.start_time }}">
+ <div class="speaker-bubble-content">{{ segment.text }}</div>
</div>
- </div>
+ {% set ns.last_speaker = segment.speaker %}
+ {% endfor %}
+ {% if transcript.segments %}</div>{% endif %}
</div>
-
+ {% else %}
<!-- Plain Text View (for non-JSON transcriptions) -->
- <div v-if="!processedTranscription.isJson" class="whitespace-pre-wrap cursor-pointer hover:bg-[var(--bg-accent-hover)] p-2 rounded transition-colors">
- ${processedTranscription.content}
- </div>
+ <div class="whitespace-pre-wrap cursor-pointer hover:bg-[var(--bg-accent-hover)] p-2 rounded transition-colors">{{ transcript.plain_text }}</div>
+ {% endif %}
</div>
</div>
@@ -365,6 +364,10 @@
const legendExpanded = ref(false);
const isDarkMode = ref(document.documentElement.classList.contains('dark'));
+ // Server-rendered transcript info (avoids duplicate JSON parsing)
+ const hasDialogue = ref({{ 'true' if transcript.has_speakers else 'false' }});
+ const plainTextTranscript = {{ transcript.plain_text|tojson|safe }};
+
// Custom audio player state
const audioIsPlaying = ref(false);
const audioCurrentTime = ref(0);
@@ -488,96 +491,8 @@
}
}
- const processedTranscription = computed(() => {
- if (!recording.value?.transcription) {
- return { hasDialogue: false, content: '', speakers: [], simpleSegments: [], bubbleRows: [] };
- }
-
- const transcription = recording.value.transcription;
- let transcriptionData;
-
- try {
- transcriptionData = JSON.parse(transcription);
- } catch (e) {
- transcriptionData = null;
- }
-
- if (transcriptionData && Array.isArray(transcriptionData)) {
- const wasDiarized = transcriptionData.some(segment => segment.speaker);
-
- if (!wasDiarized) {
- const segments = transcriptionData.map(segment => ({
- sentence: segment.sentence,
- startTime: segment.start_time,
- }));
- return {
- hasDialogue: false,
- isJson: true,
- content: segments.map(s => s.sentence).join('\n'),
- simpleSegments: segments,
- speakers: [],
- bubbleRows: []
- };
- }
-
- const speakers = [...new Set(transcriptionData.map(segment => segment.speaker).filter(Boolean))];
- const speakerColors = {};
- speakers.forEach((speaker, index) => {
- speakerColors[speaker] = `speaker-color-${(index % 8) + 1}`;
- });
-
- const simpleSegments = transcriptionData.map(segment => ({
- speakerId: segment.speaker,
- speaker: segment.speaker,
- sentence: segment.sentence,
- startTime: segment.start_time || segment.startTime,
- endTime: segment.end_time || segment.endTime,
- color: speakerColors[segment.speaker] || 'speaker-color-1'
- }));
-
- const processedSimpleSegments = [];
- let lastSpeaker = null;
- simpleSegments.forEach(segment => {
- processedSimpleSegments.push({
- ...segment,
- showSpeaker: segment.speaker !== lastSpeaker
- });
- lastSpeaker = segment.speaker;
- });
-
- const bubbleRows = [];
- let lastBubbleSpeaker = null;
- simpleSegments.forEach(segment => {
- if (bubbleRows.length === 0 || segment.speaker !== lastBubbleSpeaker) {
- bubbleRows.push({
- speaker: segment.speaker,
- color: segment.color,
- isMe: segment.speaker && (typeof segment.speaker === 'string') && segment.speaker.toLowerCase().includes('me'),
- bubbles: []
- });
- lastBubbleSpeaker = segment.speaker;
- }
- bubbleRows[bubbleRows.length - 1].bubbles.push({
- sentence: segment.sentence,
- startTime: segment.startTime || segment.start_time,
- color: segment.color
- });
- });
-
- return {
- hasDialogue: true,
- isJson: true,
- segments: simpleSegments,
- simpleSegments: processedSimpleSegments,
- bubbleRows: bubbleRows,
- speakers: speakers.map(speaker => ({
- name: speaker,
- color: speakerColors[speaker]
- }))
- };
- }
- return { hasDialogue: false, content: transcription, speakers: [], simpleSegments: [], bubbleRows: [] };
- });
+ // Note: Transcript content is now server-rendered for accessibility (Claude, screen readers, etc.)
+ // The processedTranscription computed property has been removed - content comes from Jinja2 templates
function seekAudio(startTime) {
if (recording.value.audio_deleted_at) return;
@@ -598,12 +513,8 @@
function copyTranscript(event) {
const button = event?.currentTarget;
- let textToCopy = '';
- if (processedTranscription.value.isJson) {
- textToCopy = processedTranscription.value.segments.map(s => `[${s.speaker}]: ${s.sentence}`).join('\n');
- } else {
- textToCopy = recording.value.transcription;
- }
+ // Use server-provided plain text (already formatted with speaker labels if diarized)
+ const textToCopy = plainTextTranscript || recording.value.transcription;
navigator.clipboard.writeText(textToCopy).then(() => {
animateCopyButton(button);
showToast('Transcription copied to clipboard!');
@@ -690,7 +601,7 @@
activeTab,
transcriptView,
audioPlayer,
- processedTranscription,
+ hasDialogue,
legendExpanded,
isDarkMode,
seekAudio,