Studio

Open a media file from Projects, or load one locally.

PIPELINE context + punctuation refine → → → → →

0 segments

AI Data & Model Operations Dashboard

Brimtek's workbench for multilingual speech, language, vision, model evaluation, call-center AI, media localization, conversational AI, and production delivery. The project and media selectors control the active operational context.

Core business

01 · SPEECH

Speech & Voice AI

ASR evaluation, diarization, telephony transcription, dialect and domain testing, TTS evaluation, human gold data, and custom speech model workflows.

02 · MEDIA

Media AI & Localization

Professional transcription, subtitles, dubbing, translation, live captions, media indexing, and human review for multilingual content.

03 · CONVERSATION

Conversational AI

Call-center AI, voicebots, chatbots, multilingual conversation data, response evaluation, safety review, and domain adaptation.

04 · VISION

Vision & Multimodal AI

Image and video labeling, scene and object annotation, OCR, video indexing, CV or VLM evaluation, and cross-modal quality review.

Operational coverage

Multilingualhuman data and evaluation

Regional dialectsTelephony / call-center16 kHz mediaBroadcast and captionsASR evaluationTTS evaluationChatbot / voicebot QAVision / VLM labeling

Brimtek works with a mix of client-controlled, Brimtek-created, licensed, and permitted public-source material. Rights and permitted uses are tracked by project and contract.

Current project / media

Choose a project from the top bar.

Video Indexing Lab

Scene cuts, extracted frames, transcript context, vision analysis, OCR, entities and searchable timecodes.

Coverage

No active indexed video.

Transcript / dub snapshot

No active transcript.

0

Segments labeled

0.0s

Total duration

0

Speakers tagged

0

QA warnings

-

Audio-min labeled / work-hour

-

Segments / hour

0

Eval items rated

-

Eval items / hour

Workbench entry points

Load the 15-clip Swedish speech-labeling demo set to inspect annotation, metadata, QA and export workflows.

Open Speech / Audio labeling for manual segmentation, transcription correction, tags and speaker metadata.

Re-open a native Brimtek annotation session for continued labeling or evaluation.

Domain / topic coverage (this session)

No segments yet.

Platform capabilities

Speech data & ASR operations - OpenAI diarized transcription, high-accuracy gpt-transcribe hybrid, remote WhisperX/pyannote, and imported diarization all land in one Brimtek schema.
Transcript normalization & adjudication - cached punctuation, casing, Arabic orthography and proper-name cleanup with glossary/context; raw ASR is preserved.
Versioned annotation sets - ASR, subtitle and dub groupings coexist; re-segmenting no longer destroys the previous cut.
Language data & translation - separate subtitle translation and spoken dub adaptation are generated with shared context and cached.
TTS / voice evaluation & synthesis - OpenAI expressive TTS/custom voice, Azure Speech, and remote Brimtek XTTS; delivery instructions can specify accent, tone, emotion and pacing.
Realtime speech workflows - browser audio streams through Brimtek's server-side WebSocket proxy so the OpenAI key never reaches the browser. Public capture requires HTTPS.
Computer vision / video intelligence - scene-change + periodic sampling, cached transcript context and OpenAI vision produce timestamped shot/entity/topic/on-screen-text records.

Local VAD vs. server transcription

The built-in ✨ Auto-segment in Label Audio is still a local WebAudio energy-VAD. It is useful for manual labeling because it detects speech/silence without uploading anything, but it does not transcribe or identify speakers.

For production media, use Studio → ① Transcribe. The server can now run OpenAI diarized ASR, the high-accuracy gpt-transcribe hybrid, or your remote WhisperX/pyannote service, then cache and context-refine the result before deriving subtitle or dub segment sets.

Speech / Audio Labeling

Load a clip, click ✨ Auto-segment to detect speech regions automatically, then click into any transcript box to type. Speech over 15s auto-splits.
Shortcuts: space play · S split · M merge · A auto-segment · J/K next/prev · Del remove

Source audio

Drop a .wav/.mp3 file here, or click to browse

Or load a bundled demo sample

Use current waveform selection as

Segments (0)

Selected segment

Click a segment row to edit its Start/End, Speaker, and Type here.

ASR Import / Advanced

Normal server transcription now lives in Studio and follows the active project/media automatically. Use this page only for an external GPU endpoint, Whisper/WhisperX JSON, SRT/VTT import, or other advanced/manual ingestion.

Transcription endpoint

not configured

Endpoint URL (your GPU / HF Inference Endpoint / proxy)

Auth token / API key (kept in this browser only)

Model hint (sent to your endpoint)

Language (or 'auto')

Uses the audio loaded in Label Audio.

Security: a token in a browser file is visible to anyone who opens the file. Fine for your own machine; for shared/client use, point this at a small proxy on your droplet that holds the token server-side. The button POSTs {audio, model, language} and expects Whisper-style {segments:[{start,end,text,speaker?}]} back.

Or import an existing transcript

Whisper / WhisperX JSON

Accepts {segments:[…]} with start/end/text and optional speaker.

SRT / VTT subtitles

Timecodes + text parsed into segments.

Paste raw transcript JSON

No transcript loaded yet.

Pipeline - one raw transcript, many outputs

1

Transcribe (here)

Audio → time-stamped transcript via your ASR endpoint, or import existing Whisper/SRT output.

2

Post-edit & tag → Label Audio

Transcribers correct the machine output and add non-speech/markup tags. The 50-70% margin lever: verify, don't transcribe from scratch.

3

Translate → Translate / TTS QA

Per-segment translation with length-bounds validation for TTS fit.

4

Subtitle → Export SRT/VTT

Caption-ready export with the CPS/duration readability rule applied.

5

Dub → Dubbing / Side-by-Side

Source + target text side by side per segment, with per-row playback and synth hooks.

Live CC / Live Translation

Stream microphone or browser/tab audio through the Brimtek server. The OpenAI key stays on the droplet. The active project automatically supplies source/target language and context.

No active project. Choose one from the top bar.

Mode

Expected source languages (codes)

Translation target

Latency / accuracy

Context / program description

Keywords / proper names (comma-separated)

idle

Source / live captions

Translated captions

Notes

Realtime transcription is optimized for low-latency captions and does not provide speaker labels. For final delivery, run the recorded media through the diarized file-transcription pipeline afterward. For browser/tab capture your browser may require selecting “share audio”.

Dubbing / Side-by-Side

Source ↔ target per segment, with per-row playback, merge (same-speaker guarded), split-at-second, and synth hooks. Load the media in Label Audio (or here), and rows from a CSV or the Raw Processing tab.

Source media (video/audio)

Import rows (CSV: group,speaker,start,end,source,target)

Segments (0)

Target language / direction

Translate / TTS QA

Per-segment translation review, mirroring the source → target workflow with a TTS "Now Playing" strip for reviewing synthesized voice output against the transcript.

Source language

Source direction

Target / translation language

TTS voice (for QA playback)

Translation buffer % (length-bounds rule)

Now Playing - no clip loaded

Segments (0)

No speech segments yet - add some in Label Audio first, then come back here.

Label Images

Computer-vision labeling workspace: draw bounding boxes, assign classes and attributes, and export training/evaluation annotations to COCO JSON or YOLO txt.

Drop a .jpg/.png here, or click to browse

Class for next box Attributes for next box

Annotations (0)

No boxes yet - draw on the image above.

Label Video

Video / vision-model labeling workspace: temporal actions, frame-level gesture regions, object/track metadata and validation-ready annotations for training or model evaluation.

Drop a .mp4/.webm here, or click to browse

Gesture label for next region

Track ID

Video annotations (0)

No annotations yet.

Subtitle & Dub Studio

Applied media-delivery workspace for documentary subtitling and dubbing. Works on the segments from Label Audio - QC against broadcast readability rules, build the dub script with duration budgets, export to delivery formats.

Preset

Max CPS

Max chars/line (CPL)

Max lines

Min duration (s)

Max duration (s)

Min gap between cues (ms)

Target language (dub / subs)

Run QC to check every cue against the preset.

Dub pipeline - Colab interop (parameters mirror your notebooks)

Source language

Target language

Dub utterance max (s)

Gap threshold (s)

Synthesis group max (s)

XTTS max merge gap (s)

Ref target per speaker (s)

Min slice (ms)

Import a Whisper/diarization JSON, QC it here, then export the XTTS CSV your notebook already reads.

🎙 Voice Studio - TTS & cloning

Engine

Default voice

Rate

Pitch

Auto-fit computes the speaking-rate adjustment each cue needs to land inside its time slot, and writes it into the SSML - the core problem in dubbing. Cloning requires a signed release for every voice (Contract Pack, Doc A/C).

💰 Job pricing - cost vs. Azure list

Media minutes

Azure STT $/audio-hour

Azure diarization add-on $/h

Azure TTS $/1M chars

Your cost $/audio-hour (GPU+ops)

Human QC hours per media hour

Annotator rate $/h

Your price $/media-minute

Enter job size and press Calculate.

Azure figures are placeholders - their pricing page loads values dynamically. Put your region's real numbers from the Azure pricing calculator in these fields before quoting.

Cue list & dub script (0)

No segments yet - create them in Label Audio (or import an SRT above).

Author Prompts

Write prompt-completion pairs to the Brimtek SFT schema - the 2023-style production workspace. RTL-aware, live word/char counts, per-item validation badge.

RTL 0 / 0 0 errors

No items yet - click ＋ New item, or import a .jsonl of {"prompt","completion",...} objects.

Evaluate Models

Explore a broad global model catalog, build a benchmark shortlist, then run human rubric scoring and pairwise preference judgments across language-model, speech, vision, and multimodal outputs.

Explore model catalog Hugging Face Hub first layer

Search models

Task

Availability

Sort

Checking Hugging Face connection...Catalog browsing does not run inference or spend Brimtek credits. Adding a model creates a benchmark shortlist only.

Search the Hugging Face Hub to discover public, gated, private-access, and provider-served models available to your configured account.

Benchmark shortlist

0 selected

No models selected yet.

Import items (.jsonl - one JSON per line)

Rubric item: {"id","prompt","response","language"} · Pairwise item: {"id","prompt","response_a","response_b","language"}

Or load sample items

3 rubric + 3 pairwise, multilingual - for training raters and demoing to clients.

Rater ID (attached to every rating)

Export ratings

No items loaded.

QA & Agreement

Inter-annotator agreement and gold-set scoring - the numbers that go in every client deliverable. Compare two annotators' work, or score an annotator against a gold standard.

Transcription agreement (two session JSONs)

Session A (or GOLD)

Session B (annotator under review)

Mode

Eval rating agreement (two ratings JSONLs from the Evaluate Models tab)

Ratings file - Rater 1

Ratings file - Rater 2

Reports: per-dimension exact agreement %, mean absolute difference (1-5 scale), and Cohen's κ for pairwise A/B/Tie choices. These are the statistics quoted in the Eval Pilot deliverable.

Validation Rules

Production QA gate - run all rules before delivery. Rules with no applicable data show N/A rather than failing.

-

Pass

-

Warn

-

Fail

-

N/A

Rule thresholds

Min conference SNR (dB)

VAD timing tolerance (ms)

Channel sync tolerance (ms)

Speaking rate (chars/sec, for translation bounds)

Max subtitle reading speed (CPS)

Constraint settings (transcription conventions)

Forbidden digits

Invalid characters

Min segment duration (s)

Max segment duration (s)

Words per segment - min

Words per segment - max

Min bitrate (kbps, est.)

Source language min %

Approved fillers (must carry # prefix)

File Metadata & Speaker Roster

Required file-level values and per-speaker labels, per section 4 of the transcription guidelines.

File-level values

Program / file breadcrumb

File status

Domain(s)

Topic(s) / scenario(s)

Primary language (locale code)

Primary variety

Other language(s)

Dialect / accent

Guidelines name

Addendum name

Annotator ID

Sample rate (Hz)

Channel count (multi-track meetings)

Channel offsets ms (comma-separated, one per channel - for sync verification)

Recording type

Speaker roster

Tag Reference

Full non-speech, markup, and disfluency tag set consolidated from the en_US v3.0 transcription guidelines and the IFAP 16kHz addendum.

Non-speech noise tags - human vocal

[breath]	Inhalation/exhalation between words, yawning
[cough]	Coughing, throat clearing, sneezing
[cry]	Crying / sobbing
[laugh]	Laughing, chuckling
[lipsmack]	Lipsmacks, tongue-clicks

Non-speech / non-human noise tags

[applause]	Clapping - placed exactly where it occurred (16kHz addendum)
[beep]	Beep replacing profanity or classified info
[bg-speech]	Background speech overlapping the foreground speaker
[click]	Machine or phone click
[dtmf]	Telephone keypad tone
[music]	≥1s of music/singing with no foreground speech; on-hold music
[no-speech]	Pause / silence of ≥1s, even with faint foreground noise mixed in
[noise]	Any other miscellaneous noise (screaming, rain, punching, etc.)
[ring]	Telephone ring
[sta]	Continuous static

Markup tags

<initial></initial>	Wraps an initialism/acronym spoken letter-by-letter, e.g. <initial>TV</initial>
<lang:Foreign></lang:Foreign>	Wraps a word/phrase in an unidentified non-target language
<lang:X></lang:X>	X = any capitalized language name (Arabic, Korean, Spanish, English…)
<overlap></overlap>	Wraps speech that overlaps with another speaker

Disfluency & other markup

#eh / #mm / #öh	Filler words - prefixed with #
word~	Stumbled / truncated word, trailing tilde, no space before it
(( ))	Unintelligible speech - double parentheses, no inner spaces
((word))	Best-guess / low-confidence transcription of a barely intelligible word

Rule: never insert a non-speech tag mid-word - place it immediately before the word in progress. Repeated instances of the same sound in one segment are represented only once.

Segmentation rules (quick reference)

Five primary segment types: Speech, Babble, Overlap, Music, Noise. Each segment has exactly one primary type.
Timestamps are positive floats in seconds.milliseconds (e.g. 12.345).
Trim continuous silence/white noise ≥2s from the edges of a segment - keep segments tight around the target sound.
Speech segments should not exceed 15 seconds - this workbench enforces it by auto-splitting on commit; use Split/Merge to fine-tune boundaries.
Transcription is required only for Speech (and Babble/Overlap where applicable) segments.
Single-channel files: segments with no speech or only non-speech sound on that channel should be dropped, not transcribed as [no-speech].

Export

Export the current session in the format your pipeline needs. Everything runs locally in the browser.

Transcription

Native JSON

Full schema - Type, MetaData, Speakers, Segments with TranscriptionData. Matches your delivery format. Re-importable from the Dashboard.

HF / Whisper manifest (.jsonl)

One JSON object per line: audio_filepath, text (ASR-ready), duration, language, speaker.

CSV

segment_id, start, end, duration, speaker, language, type, transcription_raw, transcription_clean.

SRT subtitles

Speech segments as SRT captions (ASR-ready text) for sync with video review tools.

WebVTT

Same as SRT but WebVTT, for HTML5 <track> / web players.

Praat TextGrid

A single IntervalTier ("segments") with all boundaries and clean text - opens directly in Praat.

Audacity label track

Tab-separated start\tend\tlabel - import via Audacity's File → Import → Labels.

Kaldi-style bundle

wav.scp, segments, text, utt2spk, spk2gender - zipped. Loads JSZip once; falls back to loose files offline.

Translation

Bilingual CSV

segment_id, start, end, source_text, target_language, translation.

Translation JSON

Same native-JSON envelope, with a Translation object added to each segment.

Full project bundle

Project folder (.zip)

/audio, /transcripts (json+csv+srt+vtt), /translations, /metadata, /exports (kaldi+textgrid+audacity) - one zip, ready to hand off.

QA summary before export

No segments yet.

Roadmap - one console, every modality

The metadata model (file-level values + annotator + typed segments) and the export pipeline are designed to extend beyond audio without a rewrite.

🖼 Photo / image labeling

Next up

Type.name: "IMAGE_ANNOTATION". Replaces time-based Segments with region-based Annotations.

Bounding boxes, polygons, and segmentation masks
Classification tags & multi-label scene/object tags
Attribute tags (occlusion, truncation, blur, low-light)
EXIF / capture metadata

🎬 Video labeling

Planned

Type.name: "VIDEO_ANNOTATION". Segments gain a spatial component per frame range.

Temporal action segments - same StartTime/EndTime pattern
Object tracking IDs (reuses SpeakerId pattern as TrackId)
Shot/scene boundary tags, camera motion tags

💬 Chatbot prompt / conversation labeling

Planned

Type.name: "CONVERSATION_ANNOTATION". Segments become conversation turns.

Turn-level role tags, intent & topic tags
Response quality rubric scoring, A/B preference ranking
Safety/policy category tags, red-team flags

🧩 Shared platform layer

Foundation - live today

What every future tab shares with the audio workbench built today:

Guidelines/AddendumName versioning per file
AnnotatorId + QA warning pipeline
Consistent native-JSON envelope
Same multi-format export layer

Projects

Each client job is a project. Upload media, run transcription on the server, then edit in Label Audio.

New project

Name

Client

Source language

Target language

All projects

Sign in to load projects.

Add media - select a project below first

Project

Fetch from URL (YouTube, Vimeo, direct link)

Quality

You are responsible for holding the rights to any media you ingest (Work Order clause 5).

…or drop a video/audio file here, or click to browse

Jobs

Server-side transcription. Auto-refreshes every 3 seconds while this tab is open.

Sign in to load jobs.

Voices & Consent

Every cloned voice must be linked to a signed release. Reference-audio upload is blocked server-side without recorded consent.

Register a voice

Voice name

Speaker label

Engine

External voice ID

Consent reference

Consent signed?

Consent date

Notes

Voice / delivery instructions (accent, tone, emotion, pacing)

Registry

Sign in to load the registry.

Users

Admin only. Roles: admin, reviewer, annotator, client.

Add user

Username

Password

Role

Initial credits

Enforce credit balance for this userExisting users remain unmetered until an admin enables enforcement.

All users and credits

Service credit rates Internal Brimtek units, configurable by admin

Loading rates...

Change my password

Current password

New password

Studio

Glossary - enforced during translation

AI Data & Model Operations Dashboard

Core business

Speech & Voice AI

Media AI & Localization

Conversational AI

Vision & Multimodal AI

Operational coverage

Current project / media

Video Indexing Lab

Transcript / dub snapshot

Workbench entry points

Domain / topic coverage (this session)

Platform capabilities

Local VAD vs. server transcription

Speech / Audio Labeling

Source audio

Segments (0)

Selected segment

ASR Import / Advanced

Transcription endpoint

Or import an existing transcript

Pipeline - one raw transcript, many outputs

Transcribe (here)

Post-edit & tag → Label Audio

Translate → Translate / TTS QA

Subtitle → Export SRT/VTT

Dub → Dubbing / Side-by-Side

Live CC / Live Translation

Source / live captions

Translated captions

Notes

Dubbing / Side-by-Side

Segments (0)

Translate / TTS QA

Segments (0)

Label Images

Annotations (0)

Label Video

Video annotations (0)

Subtitle & Dub Studio

Dub pipeline - Colab interop (parameters mirror your notebooks)

🎙 Voice Studio - TTS & cloning

💰 Job pricing - cost vs. Azure list

Cue list & dub script (0)

Author Prompts

Prompt

Completion - Response A

Item metadata

Evaluate Models

Explore model catalog Hugging Face Hub first layer

Benchmark shortlist

Rate item

Export ratings

QA & Agreement

Transcription agreement (two session JSONs)

Eval rating agreement (two ratings JSONLs from the Evaluate Models tab)

Validation Rules

Rule thresholds

Constraint settings (transcription conventions)

File Metadata & Speaker Roster

File-level values

Speaker roster

Tag Reference

Non-speech noise tags - human vocal

Non-speech / non-human noise tags

Markup tags

Disfluency & other markup

Segmentation rules (quick reference)

Export

QA summary before export

Roadmap - one console, every modality

🖼 Photo / image labeling

🎬 Video labeling

💬 Chatbot prompt / conversation labeling

🧩 Shared platform layer

Projects

New project

All projects