Find the lab that fits your work.

The problem

You mix your own tracks. You want a sober second opinion before you bounce (loudness target, harshness, stereo width) without paying for a mastering engineer on every demo.

Our recommendation

MixLab Analyzer for pre-bounce QA. Once your track passes the sniff test, use any AI mastering tool to render. Start with the Free tier; upgrade to Plus when you want every analysis preset, ten times the API volume, custom targets and deep analysis on every upload.

Relevant labs

• MixLab
• SkillLab

Where to start

• Open MixLab Analyzer
• AI mastering tools compared
• Practice with SkillLab

The problem

You record yourself and a guest. You want consistent loudness, no filler-heavy stretches, no surprises in the pauses, before you ship the episode and the listener disengages.

Our recommendation

VoiceLab QA for per-episode quality check. Aim for -16 LUFS integrated, SNR above 28 dB, pauses that read as rhythm rather than hesitation, no clipping. Drop a whole folder of episodes into the workbench when you want to see them side by side.

Relevant labs

• VoiceLab
• SignalLab

Where to start

• Try VoiceLab QA
• Pragmatic podcast loudness
• What envelope decay after speech actually measures

The problem

You can hear that a take drifted, but not where or by how much. A coach hears it once; you want to see the note over time, jump straight to the stretch that slid, and know whether your vibrato is a wobble or a rate.

Our recommendation

The singing tool in VoiceLab follows the note instead of judging the room: pitch over time with the drifting stretch playable back, vibrato rate and depth, and the range you actually held steadily rather than the one note you hit once. It says so when a take was not sung, instead of inventing a range for speech.

Relevant labs

• VoiceLab
• SkillLab

Where to start

• Analyse a sung take
• Speech QA on the same take
• Practise with SkillLab

The problem

You want to log your hearing experience across the day (restaurants, meetings, TV) and share a structured summary with your audiologist. Captions when you need them. Routing visibility when it matters.

Our recommendation

HearLab Companion. Non-medical by design: it logs what you experience, not what your hearing measures. Use it alongside your audiological care, not instead of it.

Relevant labs

• HearLab

Where to start

• Try HearLab Companion
• Why non-medical framing matters
• Live Caption on Android: what it does

The problem

You ingest thousands of audio files per day. You need to classify, tag, QA-gate, and search them at scale. The MVP needs to ship in two weeks, not two quarters.

Our recommendation

SignalLab Indexer API: drop in, get structured JSON out. Content classification, region timeline, QA flags, machine-readable tags. Studio tier for production usage and the orchestration UI.

Relevant labs

• SignalLab

Where to start

• Try SignalLab Indexer
• See the API endpoints
• A useful tag schema for archives

The problem

Your audio sometimes routes to the wrong destination. You forget to mute the browser source. The local recording sometimes isn’t rolling. You want one pane of glass for routing + checklist + health.

Our recommendation

CueLab Monitor. Pre-show checklist, routing graph, live health checks. Catches the things you forget when you’re thirty seconds from going live.

Relevant labs

• CueLab

Where to start

• Try CueLab Monitor
• OBS routing without surprises
• A reusable pre-show checklist

The problem

You want your students to practise audio skills with feedback that lands. The available apps are either toys (gamified slop) or industrial (DAW-only, expensive). You need something in between.

Our recommendation

SkillLab Challenges. Tiered exercises with real-feedback scoring. Progress survives a missed day. Honest about what it does and doesn’t teach. Studio tier for cohort use.

Relevant labs

• SkillLab
• VoiceLab

Where to start

• Try SkillLab Challenges
• Designing tiered audio challenges
• Feedback patterns from real coaches

The problem

You produce voice content in multiple languages. You need to QA pacing, clarity, room consistency, and pronunciation risk across hundreds of takes from different voice actors.

Our recommendation

VoiceLab QA at scale. Per-vertical thresholds (e.g. consumer e-learning vs broadcast). Studio tier for shared workspaces and integration with your existing TMS or content pipeline.

Relevant labs

• VoiceLab
• SignalLab

Where to start

• Try VoiceLab QA
• What envelope decay after speech actually measures
• See VoiceLab API endpoint

The problem

You’re building an audio app. You want to understand the engineering choices, integrate analysis primitives into your stack, and read deep technical material that isn’t marketing fluff.

Our recommendation

Docs, insights, and the AudioLab API. Everything we ship comes with a build log. Use what we built as primitives or learn how we built it.

Relevant labs

• MixLab
• SignalLab
• VoiceLab
• HearLab
• CueLab
• SkillLab

Where to start

• All engineering insights
• Building MixLab (the build log)
• WebAudio vs native, the line in 2026
• The hosted API

The problem

You test voice agents or phone calls and need audio facts next to the transcript: was the line too quiet, clipped, noisy or full of silence, and did the synthetic voice leave the platform at a sane loudness? A transcript alone does not show any of that.

Our recommendation

Put the audio analysis behind one key: loudness, true peak and clipping, voice quality (SNR, pauses, sibilance) and a speech, music, noise and silence timeline come back as JSON, from the hosted API or over MCP inside your agent. The loudness engine is checked against the EBU Tech 3341/3342 signals and every case is published. It is EU-hosted, an uploaded file is swept within the hour, and the free tier is 1,000 calls and 2 GB a month. It measures the audio; it does not judge what was said.

Relevant labs

• VoiceLab
• SignalLab

Where to start

• Try the Voice AI check on one recording
• See the API and its published checks
• Connect an agent over MCP
• Read the methodology

The problem

You need a first answer to “is it loud enough here to matter?” before anyone commissions an exposure assessment: the level over time, the peaks, and how long a working day at this level takes to reach the 85 dB(A) upper action value.

Our recommendation

NoiseLab on Android: A, C or Z weighting to IEC 61672, LAeq, L10 and L90, peaks, and the hours you can stay before 85 dB(A), with a plain note while you are under the 80 dB(A) lower action value. It is a first scan: a phone is not a class 1 or class 2 sound level meter, the app shows < 30 or > 105 instead of a number outside what a phone microphone can measure, and every screen says the reading is uncalibrated until you calibrate it against a real meter. A formal exposure assessment still needs a competent person. Pro adds the share of the session above the 80 and 85 dB(A) action values and puts the numbers in a report you can hand over.

Where to start

• NoiseLab for Android
• How loud is too loud: a dB(A) table with the hours you can stay

The problem

You need to know whether a room talks back before money goes into it: a classroom, a meeting room, a hall or a rehearsal space. Reverberation time per octave band is the number that tells, and a first read should not need a rented rig.

Our recommendation

RoomLab on Android: reverberation time per octave band from a clap or a balloon pop, flutter echo and the noise floor, a score out of 100 for what the room is for (classroom, office, lecture hall, speech, music, recording or a gathering) with the source of every limit shown, and a map of the room from a few spots. A clap reads from 500 Hz up and a speaker sweep covers the low bands. It is a quick check, not an ISO 3382 measurement: use it to decide where a proper measurement is worth it.

Where to start

• RoomLab for Android
• How to measure RT60 with your phone