Speakr is a personal, self-hosted web application designed for transcribing audio recordings
Python
docs+i18n: document exact speaker count, threaded workers, chat timestamp chips and new connectors; agentic Inquire gains concrete example queries in README and user guide; fix stale 11-language claims; localize the empty-state landing panel and the No Folder option
Murtaza Nasirc8206ccb6b09
diff --git a/README.md b/README.mdindex c4e3cf24..50205740 100644--- a/README.md+++ b/README.md@@ -40,18 +40,18 @@ Speakr turns a recording into organized, searchable, shareable knowledge. Here i- **Hands-off intake** - a watched "black hole" folder auto-imports and processes any audio dropped into it, with optional per-user subfolders named by user id or username.### Transcribe-- **Bring your own engine** - self-hosted WhisperX (recommended; it is what enables the speaker features below), OpenAI, Mistral / Voxtral, AssemblyAI, or any custom ASR webservice. The right connector is auto-detected from your configuration.+- **Bring your own engine** - self-hosted WhisperX (recommended; it is what enables the speaker features below), OpenAI, Mistral / Voxtral, AssemblyAI, OpenASR, Alibaba FunASR, or any custom ASR webservice. The right connector is auto-detected from your configuration.- **Speaker diarization** - automatic who-said-what labeling (WhisperX, or OpenAI's diarizing models).- **Voice profiles** - recognize the same person across different recordings via voice embeddings (requires the WhisperX ASR backend).- **Custom vocabulary and hotwords** (most effective with the WhisperX backend) - bias the transcriber toward names, jargon, and acronyms it would otherwise mishear; configurable globally, per tag / folder, or right in the upload and in-app recording dialogs, with reusable saved templates.- **Synced playback** - click any line to jump to that moment, follow-along highlighting during playback, and a chat-style bubble view.-- **Language support** - automatic language detection plus a quick-pick of 11 common languages.+- **Language support** - automatic language detection plus the full Whisper language list (about one hundred languages), with names localized to your interface language.### Understand- **Summaries** - generated automatically, with prompts you can fully customize per recording, tag, or folder (including reusable prompt variables).- **Event extraction** - surface action items and calendar-worthy events from a transcript.-- **Per-recording chat** - ask questions about a single recording in a floating, dockable panel.-- **Inquire Mode** - semantic search and natural-language chat across your entire library at once. An optional **agentic mode** (opt-in beta) upgrades this to an AI researcher that iteratively searches, lists, and reads your recordings until it can answer - showing every step live, and citing the exact moment with numbered source links that open the recording and play from that point. You decide what it may read: transcripts always, summaries and private notes only if you allow them.+- **Per-recording chat** - ask questions about a single recording in a floating, dockable panel; answers cite transcript timestamps as clickable chips that start playback at that moment.+- **Inquire Mode** - semantic search and natural-language chat across your entire library at once. An optional **agentic mode** (opt-in beta) upgrades this to an AI researcher that iteratively searches, lists, and reads your recordings until it can answer - showing every step live, and citing the exact moment with numbered source links that open the recording and play from that point. Ask "what did we decide about the pricing change, and did anyone disagree?" and it reads across the relevant meetings and cites each position to the second it was said. You decide what it may read: transcripts always, summaries and private notes only if you allow them.### Organize- **Folders and bulk operations** to keep a large library tidy.@@ -151,6 +151,8 @@ Speakr uses a **connector-based architecture** that auto-detects your transcript| **VibeVoice ASR** | Self-hosted (vLLM) | Yes (built-in) | No || **MOSI/Mossland MOSS** | Hosted API key | Yes (built-in) | No || **AssemblyAI** | Just API key | Yes (built-in) | No |+| **OpenASR** | Self-hosted server | Yes (built-in) | No |+| **Alibaba FunASR** | API key + OSS bucket | Yes (built-in) | No || **Legacy Whisper** | Just API key | No | No |**Simplest setup (OpenAI with diarization):**diff --git a/docs/features.md b/docs/features.mdindex 09ba1d1c..f53d3265 100644--- a/docs/features.md+++ b/docs/features.md@@ -30,6 +30,8 @@ Speakr supports multiple transcription engines to match your needs and budget. UWhen using the [ASR endpoint](getting-started.md#option-b-custom-asr-endpoint-configuration), Speakr automatically identifies different speakers in your recordings. If you encounter issues, check the [troubleshooting guide](troubleshooting.md#speaker-identification-not-working). Each speaker gets a unique label that you can later customize with actual names. The system remembers these speaker profiles, building a library that improves identification accuracy over time. Manage your speaker library in [account settings](user-guide/settings.md). This feature transforms multi-person meetings from walls of text into organized conversations.+You can guide diarization with a speaker count. The upload, recording, and reprocess dialogs offer a Range / Exact toggle (remembered per user): Range sends a minimum and maximum, while Exact sends one number, and providers that only accept an exact count (such as OpenASR) show the single field automatically. Tag and folder defaults work with either form.+**Cloud Diarization with OpenAI**: When using OpenAI's `gpt-4o-transcribe-diarize` model, speaker diarization is also available with speakers labeled as A, B, C, etc. For longer files (over ~23 minutes) that require chunking, the system maintains speaker identity across chunks using audio reference samples. This technique supports **up to 4 speakers** - recordings with more speakers may have inconsistent labels across different sections of the transcript.**Cloud Diarization with Mistral**: Mistral's Voxtral model provides built-in speaker diarization and automatic language detection. It handles chunking internally, so long files work without extra configuration.diff --git a/docs/index.md b/docs/index.mdindex f29828fc..52b63ed0 100644--- a/docs/index.md+++ b/docs/index.md@@ -72,7 +72,7 @@ Speakr takes a recording from raw audio to organized, searchable, shareable know<ul><li><a href="features#multi-engine-support">Bring your own engine: WhisperX, OpenAI, Mistral, custom ASR</a></li><li><a href="features#speaker-diarization">Speaker diarization</a> and <a href="features#speaker-management">voice profiles</a> (WhisperX backend)</li>- <li><a href="features#language-support">Auto-detect plus 11 common languages</a></li>+ <li><a href="features#language-support">Auto-detect plus the full Whisper language list</a></li><li>Custom vocabulary and hotwords (most effective with WhisperX)</li></ul></div>diff --git a/docs/troubleshooting.md b/docs/troubleshooting.mdindex a9089307..5ffbb7ff 100644--- a/docs/troubleshooting.md+++ b/docs/troubleshooting.md@@ -143,6 +143,10 @@ If the interface shows blank panels or raw translation keys (console messages liFrom v0.10.3-alpha onward this heals itself: each release installs a service worker with its own cache namespace, static assets are revalidated on load, and translation files are always fetched fresh, so an image upgrade can no longer strand a stale app shell. Upgrades from versions before v0.10.3-alpha may still need the one-time manual clear described above.+### App Unresponsive While Media Plays or Chats Stream++Since v0.10.4-alpha the stock image runs threaded gunicorn workers (3 processes x 8 threads), so streaming responses such as chat answers and audio or video playback no longer occupy one of only three request slots. If you override the container command, adopt `--worker-class gthread --threads 8` to get the same behavior. Behind nginx, `proxy_buffering on` helps further: nginx absorbs each response at local speed and feeds slow clients itself, freeing the application thread in seconds rather than for the duration of playback.+## Feature-Specific Issues### Chat Responses End Mid-Sentencediff --git a/docs/user-guide/inquire-mode.md b/docs/user-guide/inquire-mode.mdindex ae25bd38..58ffbea2 100644--- a/docs/user-guide/inquire-mode.md+++ b/docs/user-guide/inquire-mode.md@@ -84,7 +84,9 @@ Performance scales well with proper setup. The initial indexing of recordings ha## Agentic Inquire (Beta)-Administrators can enable an agentic mode with `ENABLE_INQUIRE_AGENT=true` (requires Inquire Mode to be enabled). Instead of a single retrieval pass, an AI agent works iteratively: it searches your transcripts, lists recordings, and reads transcripts, summaries, or notes as needed, thinking between steps until it can answer. The interface shows each step live (what was searched, what was read, how many matches), a Stop button to interrupt, and a collapsed activity line on every answer so you can always see how a conclusion was reached. Answers cite their sources as links back to the recordings they came from.+Administrators can enable an agentic mode with `ENABLE_INQUIRE_AGENT=true` (requires Inquire Mode to be enabled). Instead of a single retrieval pass, an AI agent works iteratively: it searches your transcripts, lists recordings, and reads transcripts, summaries, or notes as needed, thinking between steps until it can answer. The interface shows each step live (what was searched, what was read, how many matches), a Stop button to interrupt, and a collapsed activity line on every answer so you can always see how a conclusion was reached. Answers cite their sources as compact numbered markers such as "2 @ 14:05": clicking one opens that recording and starts playback at the cited moment, and a Sources list at the end of each answer maps every number to its recording. The numbering stays stable through the conversation, so a follow-up like "tell me more about 3" refers to exactly what you saw.++A few examples show the difference from classic Inquire. Ask "what did we decide about the pricing change, and did anyone disagree?" and the agent will search for the pricing discussions, read the relevant transcript sections across several meetings, and answer with each position cited to the moment it was said. Ask "summarize everything client X asked for this quarter" and it will list the recordings involving that client, read their summaries, and consolidate the requests with one citation per item. Ask "when did I first mention the migration plan?" and it will search, compare dates, and hand you a link that starts playback at that sentence. Classic Inquire retrieves passages in one pass and answers from whatever that single search surfaced; the agent keeps working until the question is actually answered, which is the difference you feel on questions that span recordings or need a fact checked rather than paraphrased.The agent respects your privacy settings. Under Account, the Inquire Privacy section controls what the agent may read: transcripts are always available (they are the minimum for Inquire to function), summaries are optional and on by default, and your notes are optional and off by default. A disabled content type is removed from the agent's toolbox entirely. Only your own notes are ever used, never notes on recordings shared with you. The active filter selection is a hard boundary: the agent can narrow its search within your filters but can never reach recordings outside them.diff --git a/static/locales/de.json b/static/locales/de.jsonindex 9c2c3757..6de7ca59 100644--- a/static/locales/de.json+++ b/static/locales/de.json@@ -1987,5 +1987,12 @@"warningSummary": "Dadurch werden alle manuellen Änderungen an Titeln und Zusammenfassungen überschrieben.","warningTranscription": "Dadurch werden alle Transkriptionen, Titel und Zusammenfassungen überschrieben. Manuelle Änderungen gehen verloren.","whatToReprocess": "Was möchten Sie neu verarbeiten?"+ },+ "emptyState": {+ "readyToCapture": "Bereit zur Aufnahme",+ "readyBody": "Ziehen Sie eine Audiodatei hierher, nehmen Sie über das Mikrofon auf oder erfassen Sie Systemaudio. Speakr erstellt daraus ein durchsuchbares Transkript mit Zusammenfassung und Chat.",+ "newRecording": "Neue Aufnahme",+ "selectRecording": "Aufnahme auswählen",+ "selectBody": "Wählen Sie eine aus der Seitenleiste oder starten Sie eine neue Aufnahme."}}diff --git a/static/locales/en.json b/static/locales/en.jsonindex c023926d..b5da4256 100644--- a/static/locales/en.json+++ b/static/locales/en.json@@ -1987,5 +1987,12 @@"warningSummary": "This will overwrite any manual edits to titles and summaries.","warningTranscription": "This will overwrite all transcriptions, titles, and summaries. Manual edits will be lost.","whatToReprocess": "What do you want to reprocess?"+ },+ "emptyState": {+ "readyToCapture": "Ready to capture",+ "readyBody": "Drop an audio file, record from your microphone, or capture system audio. Speakr turns it into a searchable transcript with summary and chat.",+ "newRecording": "New recording",+ "selectRecording": "Select a recording",+ "selectBody": "Pick one from the sidebar or start a new recording."}}diff --git a/static/locales/es.json b/static/locales/es.jsonindex f4de793e..ad388e1a 100644--- a/static/locales/es.json+++ b/static/locales/es.json@@ -1987,5 +1987,12 @@"warningSummary": "Esto sobrescribirá cualquier edición manual de títulos y resúmenes.","warningTranscription": "Esto sobrescribirá todas las transcripciones, títulos y resúmenes. Las ediciones manuales se perderán.","whatToReprocess": "¿Qué desea reprocesar?"+ },+ "emptyState": {+ "readyToCapture": "Listo para grabar",+ "readyBody": "Arrastre un archivo de audio, grabe desde el micrófono o capture el audio del sistema. Speakr lo convierte en una transcripción con resumen y chat.",+ "newRecording": "Nueva grabación",+ "selectRecording": "Seleccione una grabación",+ "selectBody": "Elija una en la barra lateral o inicie una nueva grabación."}}diff --git a/static/locales/fr.json b/static/locales/fr.jsonindex 4dda7864..0b1d82db 100644--- a/static/locales/fr.json+++ b/static/locales/fr.json@@ -1987,5 +1987,12 @@"warningSummary": "Cela écrasera toutes les modifications manuelles des titres et des résumés.","warningTranscription": "Cela écrasera toutes les transcriptions, tous les titres et tous les résumés. Les modifications manuelles seront perdues.","whatToReprocess": "Que souhaitez-vous retraiter ?"+ },+ "emptyState": {+ "readyToCapture": "Prêt à enregistrer",+ "readyBody": "Déposez un fichier audio, enregistrez depuis le microphone ou capturez l'audio système. Speakr le transforme en transcription consultable avec résumé et chat.",+ "newRecording": "Nouvel enregistrement",+ "selectRecording": "Sélectionnez un enregistrement",+ "selectBody": "Choisissez-en un dans la barre latérale ou lancez un nouvel enregistrement."}}diff --git a/static/locales/pt-BR.json b/static/locales/pt-BR.jsonindex 7d05fbad..4e510de3 100644--- a/static/locales/pt-BR.json+++ b/static/locales/pt-BR.json@@ -1987,5 +1987,12 @@"warningSummary": "Isso substituirá todas as edições manuais de títulos e resumos.","warningTranscription": "Isso substituirá todas as transcrições, títulos e resumos. As edições manuais serão perdidas.","whatToReprocess": "O que você deseja reprocessar?"+ },+ "emptyState": {+ "readyToCapture": "Pronto para gravar",+ "readyBody": "Arraste um arquivo de áudio, grave pelo microfone ou capture o áudio do sistema. O Speakr o transforma em uma transcrição pesquisável com resumo e chat.",+ "newRecording": "Nova gravação",+ "selectRecording": "Selecione uma gravação",+ "selectBody": "Escolha uma na barra lateral ou inicie uma nova gravação."}}diff --git a/static/locales/ru.json b/static/locales/ru.jsonindex 10ccbb71..b021398a 100644--- a/static/locales/ru.json+++ b/static/locales/ru.json@@ -1987,5 +1987,12 @@"warningSummary": "Все ручные правки заголовков и резюме будут перезаписаны.","warningTranscription": "Все транскрипции, заголовки и резюме будут перезаписаны. Ручные правки будут потеряны.","whatToReprocess": "Что вы хотите обработать заново?"+ },+ "emptyState": {+ "readyToCapture": "Готово к записи",+ "readyBody": "Перетащите аудиофайл, запишите с микрофона или захватите системный звук. Speakr превратит это в транскрипт с поиском, сводкой и чатом.",+ "newRecording": "Новая запись",+ "selectRecording": "Выберите запись",+ "selectBody": "Выберите запись на боковой панели или начните новую."}}diff --git a/static/locales/zh.json b/static/locales/zh.jsonindex f31b85aa..adb64495 100644--- a/static/locales/zh.json+++ b/static/locales/zh.json@@ -1987,5 +1987,12 @@"warningSummary": "这将覆盖对标题和摘要的所有手动编辑。","warningTranscription": "这将覆盖所有转录、标题和摘要。手动编辑将丢失。","whatToReprocess": "您要重新处理什么?"+ },+ "emptyState": {+ "readyToCapture": "准备录制",+ "readyBody": "拖入音频文件、通过麦克风录制或捕获系统音频。Speakr 会将其转换为可搜索的转录文本,并附带摘要和聊天。",+ "newRecording": "新建录音",+ "selectRecording": "选择一个录音",+ "selectBody": "从侧边栏中选择一个,或开始新的录音。"}}diff --git a/templates/components/detail/empty-state.html b/templates/components/detail/empty-state.htmlindex 44a8809d..aa38cfb5 100644--- a/templates/components/detail/empty-state.html+++ b/templates/components/detail/empty-state.html@@ -27,15 +27,11 @@style="background: var(--bg-tertiary);"><i class="fas fa-microphone text-3xl" style="color: var(--text-accent);"></i></div>- <h2 class="text-2xl font-semibold mb-2" style="color: var(--text-primary);">- Ready to capture- </h2>- <p class="t-body mb-6" style="color: var(--text-muted);">- Drop an audio file, record from your microphone, or capture system audio — Speakr turns it into a searchable transcript with summary and chat.- </p>+ <h2 class="text-2xl font-semibold mb-2" style="color: var(--text-primary);" v-text="t('emptyState.readyToCapture')"></h2>+ <p class="t-body mb-6" style="color: var(--text-muted);" v-text="t('emptyState.readyBody')"></p><div class="flex flex-col sm:flex-row gap-2 justify-center"><button @click="switchToUploadView" class="btn btn--primary">- <i class="fas fa-plus"></i><span>New recording</span>+ <i class="fas fa-plus"></i><span v-text="t('emptyState.newRecording')"></span></button></div></template>@@ -46,14 +42,10 @@style="background: var(--bg-tertiary);"><i class="fas fa-list text-xl" style="color: var(--text-muted);"></i></div>- <h2 class="text-lg font-semibold mb-1" style="color: var(--text-primary);">- Select a recording- </h2>- <p class="t-meta mb-5">- Pick one from the sidebar or start a new recording.- </p>+ <h2 class="text-lg font-semibold mb-1" style="color: var(--text-primary);" v-text="t('emptyState.selectRecording')"></h2>+ <p class="t-meta mb-5" v-text="t('emptyState.selectBody')"></p><button @click="switchToUploadView" class="btn btn--ghost">- <i class="fas fa-plus"></i><span>New recording</span>+ <i class="fas fa-plus"></i><span v-text="t('emptyState.newRecording')"></span></button></template></div>diff --git a/templates/components/shared/folder-tag-picker.html b/templates/components/shared/folder-tag-picker.htmlindex 85dc9321..60b0bc11 100644--- a/templates/components/shared/folder-tag-picker.html+++ b/templates/components/shared/folder-tag-picker.html@@ -23,7 +23,7 @@<div class="relative"><select v-model="selectedFolderId"class="w-full pl-8 pr-8 py-2 text-sm rounded-lg cursor-pointer focus:outline-none focus:ring-2 focus:ring-[var(--border-focus)] appearance-none border border-[var(--border-secondary)] bg-[var(--bg-tertiary)] text-[var(--text-primary)] hover:bg-[var(--bg-input)] transition-colors">- <option :value="null">No Folder</option>+ <option :value="null" v-text="t('folderManagement.noFolder')"></option><option v-for="folder in availableFolders" :key="folder.id" :value="folder.id">${ folder.name }</option>