NEWS

Echoteka news

Releases and updates, newest first. Full list of changes: changelog. Follow news on Telegram.

Poll results: pitch correction first, then backing vocals

Our first poll, "What do we build next?", closes today. Thank you to everyone who voted. The count at 12:30 today: gentle vocal pitch correction 12 votes, backing vocals and vocal doubles 6, EQ in plain words 4, one-sound album mastering 2, synced lyrics 1, an iPhone app 0. Signed-in users could vote, one vote each.

You picked two tools, and both are already built.

  • Gentle vocal pitch correction is live now (open it). The voice is pulled toward the nearest note, or toward the steps of the track's key. Vibrato and slides stay, and a correction never exceeds one semitone. Two presets, "Gentle" and "Noticeable", plus a strength control. Upload a vocal track or a finished song. The first 30 seconds are a free preview with before/after; the full version needs a sign-in and costs 1 credit per run, for a recording up to 5 minutes. If processing fails on our side, the credit comes back.
  • Backing vocals and doubles come next. A double of the vocal, or a harmony: a third up, a third down or a fifth up, in the key of the track. You get a separate backing track and a mix with the vocal. Harmonies need extra setup on the server, so this tool follows pitch correction.

The next poll opens in the app soon, for signed-in users.

Also new since November 9:

  • Analysis progress in one place. The AI advice block now shows the whole path: the upload with a percentage, the analysis of loudness, peaks and dynamics, and the preparation of the advice. The start button no longer shows a stuck "0%".
  • A more compact upload box on phones. The icon sits on the left and the hint takes one line.
  • The Master button no longer stays silent. When the app asks a question before the run, for example about a loud or quiet input, the button now scrolls to the question, highlights it and says to pick an option first.

Long tracks and audiobooks up to 2 hours

Echoteka now masters long recordings: audiobooks, podcasts, lectures, DJ sets and live shows up to 2 hours in one file. It is part of mastering in the app.

  • Price before you start. Upload a file longer than 10 minutes and the app shows the price and the processing time first. Up to 10 minutes costs 1 credit, like regular mastering; every started 15 minutes after that adds 1 credit. For example, 47 minutes is 4 credits and 2 hours is 9 credits.
  • Four types. Voice and Podcast at −16 LUFS, Audiobook at −19 LUFS with peaks no higher than −3 dBTP (the ACX guide), and Music at −14 LUFS for DJ sets and live shows.
  • No loudness jumps. Loudness, tone and the noise profile are set from the whole file, then the recording is processed in one-minute parts and joined with crossfades.
  • Close the tab. Processing runs on our server. We email you when it is ready, and the result stays in "My long tracks" for 72 hours.
  • Credits come back if processing fails or you cancel it.
  • Formats: WAV 24-bit, FLAC or MP3 out. Long recordings need a free account.

How it works, step by step: master a long audiobook or podcast in one file.

Echoteka 1.6: a new look, long tracks, audio to text, refine with AI

Echoteka 1.6 brings a new look of the app and tools for long recordings and speech. All of it is live in the app.

  • A new look: tiles and a Pro panel. Tools and settings are compact tiles with an icon, a word or two and a short value. Fine tuning (sliders, the processing chain, presets) folds into the Pro panel. Four themes: Signal, Daylight, Soft and High contrast, plus System. Everything from before is still there: the progress ring, the track passport, the before/after player, variants and formats.
  • Long tracks and audiobooks up to 2 hours. Audiobooks, podcasts, lectures, DJ sets and live shows in one file. Up to 10 minutes costs 1 credit, and every started 15 minutes adds 1 credit: 47 minutes is 4 credits, 2 hours is 9 credits. You see the price first, you can close the tab, we email you when it is ready, and credits come back if processing fails. How it works.
  • Audio to text. Transcripts with timecodes and a player, TXT, DOCX, SRT and VTT subtitles, and a summary with topics and action items. English, German, French, Spanish, and Bulgarian in beta, with auto-detection. 30 free minutes a day when signed in, 5 without an account, then 1 credit per started hour. Open it.
  • Refine with AI advice. On the result screen the AI turns its advice into settings for a new master: style, loudness, bass and treble. You see what will change and the price before the run. Available after sign-in.
  • Smart speech cleanup in the audio cutter. It finds "um" and "uh", breaths, mouth clicks and level jumps and marks them on the waveform, and you decide what to cut. 5 free minutes a day, then 1 credit per started 10 minutes.
  • AI artifact cleanup and mastering in one run, now with your own mastering settings and no second upload.

Guests can master a track for free every day as MP3; WAV and FLAC come with a free account.

New in Echoteka: audio to text, with SRT subtitles

Echoteka can now turn speech into text. Upload a recording of a meeting, interview, lecture, podcast, voice note or video, and get the text back with timecodes. It is live in the app.

  • Text with timecodes and a player. Click a word to hear that moment; the current word is highlighted while it plays. Search and edit right on the page.
  • Downloads: TXT with or without timecodes, DOCX, and SRT or VTT subtitles for YouTube and video editors.
  • In brief, topics and action items right after recognition: three lines with the essence, topics with timecodes and who does what by when. After sign-in, other formats: meeting, interview, lecture, podcast or your own request.
  • Languages: English, German, French, Spanish, and Bulgarian in beta, with auto-detection.
  • Free every day: 30 minutes when signed in, 5 minutes without an account. Later, beyond the free daily minutes, 1 credit per started hour.
  • Privacy: the audio is deleted right after transcription, the text is kept for 30 days, and recordings are not used to train models.

Guides: audio to text, how to make SRT subtitles from a video for free, how to transcribe a meeting or an interview and voice message to text.

Echoteka 1.5: a clearer mastering screen, honest cleanup, more tools

Echoteka 1.5 is about knowing what will happen to your track before you press the button, and what happened to it afterwards. All of this is live in the app.

  • An AI advice right after upload. The track studio suggests a style and a loudness with the reason in one sentence, and one button starts the run from that advice.
  • A new mastering screen. It opens with "Auto will do: style and loudness", so nothing is a surprise. Four sound variants appear on their own after the analysis, with matched loudness, and a tap picks the style (signed-in users get 30 previews a day). "Soften harshness" and "Louder and denser" are one tap away. Choose WAV or another format and 44.1, 48 or 96 kHz. On a desktop the settings sit in two columns and one "Master" button stays in view. How to choose a style before you master.
  • Processing you can follow. Upload and processing are shown separately, with a time estimate, seven named steps ("Now: ..."), the waveform of your track and a "Finalizing" state on the last percent. After a site update an open tab offers to refresh itself.
  • A result that starts with a summary. What was done, the style, loudness before and after, the format and the sample rate. The main button reads "Download WAV 16-bit 44.1 kHz". "Not right? Try differently" gives a Spotify or Apple Music version, louder or quieter, more bass or more top. Two charts, "Loudness over time" and "Spectrum before/after", and one "Original / Master" player with matched loudness and a loop: select a part and listen on repeat.
  • More air, less dullness. In the Universal and Trap styles the assistant's tonal correction now switches on only when a defect is found. We checked it on 159 tracks; loudness and peaks do not change. The share of tracks with an air cut of 1 dB or more fell from about half to about 10%.
  • Honest AI artifact cleanup. Before the run it says plainly that the cleanup weakens audible generator marks and hiss, and that it does not promise detectors will not find the track. If no marks are found, your credit comes back and you can master the track at once. "Download cleaned WAV" and "Master the cleaned track" are on top, the player switches between Original and Cleaned, and you can come back to the task within 24 hours after closing the tab. In the studio there is one run that cleans and masters.
  • Vocal removal results. "Download instrumental" and "Download vocal" on top, an "Original / Instrumental / Vocal" player, "Listen to 30 s" before processing ends, and "Tell me when it is ready" by email. "Put a voice on this instrumental" opens with the instrumental already loaded. See the vocal remover.
  • Tools back inside the app. Album mastering for several tracks with one loudness and sound; stem mastering for 2 to 8 stems of one track; the mastering calculator and the loudness penalty with listening after the platform's normalization; audio restoration with cassette, vinyl, phone and interview sets and fine tuning.
  • A processing chain. "More settings" has modules, strength, your own presets and community presets. "More about the track" shows the spectrum, stereo and phase and how each platform will turn the track down, before you run anything.
  • Convenience. Every format the server understands is accepted: wav, mp3, flac, m4a, ogg, aac, aiff, aif, webm, opus, amr and oga. Install the app on a phone or a computer. The account page has password change, API keys, history export to CSV and account deletion, plus "Invite a friend" and achievements. "What's new" is on the home page, and the in-app assistant can see the track you have open.
  • Vote on what we build next. The weekly question "What do we build next?" is open to signed-in users, one vote each, and it is extended to November 15. The wishes form now shows "It can already do this" when you ask for something that exists.

Short on time? Start with the style preview guide.

Echoteka 1.4: before and after in every result, faster masters, vintage sound

A few changes went live since the last update, and all of them are in the app now.

  • Before and after in every result. A mastering result opens with a before/after player. Switch between the original and the master with the buttons or the keys 1 and 2, and the volume is matched by default, so you hear the difference in the sound, not in the level. Open the app to try it.
  • Faster masters. The processing core is about 21% faster in our benchmark on reference tracks. The output is bit-identical in all 20 checks. The benchmark covers the processing itself, so the gain on your file depends on its style and length.
  • Loud tracks stay loud. If a track is already loud and you have not set a loudness target yourself, the master is not turned down below −10 LUFS.
  • Vintage sound. Vintage sound gives your song a 60s vinyl, 70s tape, 80s cassette, AM radio or lo-fi character, with an amount control and a free 25-second preview. The lo-fi converter is there for chill and moody versions.
  • Music visualizer. Music visualizer makes a beat-synced video with your cover for TikTok, Reels and Shorts, in 9:16, 1:1 or 16:9.
  • An assistant that knows the tools. The in-app assistant answers from a knowledge base of every Echoteka tool and points you to the right one for your task. It passed an internal exam with 24 answers out of 24.
  • Track studio. The app is now a studio for one track: the before/after player, goals, steps and versions on one screen, and Goals, Step and Versions tabs on a phone. Every result is saved as a version you can compare and download.
  • What next. A finished master suggests the next step: a clip, a track check or a release pack. After vocal removal you can change the key or record your voice over the instrumental.
  • Progress you can trust. The wave circle with a percentage is back, uploads show the real percentage sent, and the queue no longer freezes at one number.
  • Quick mastering profiles. Spotify / YouTube, Apple Music, Club, Vintage and Podcast set the style and platform in one tap; the sound processing does not change.
  • Clear credit prices. The pricing page shows what each tool costs in credits and how long a track can be.
  • Karaoke guide. Karaoke at home for beginners.
  • Spanish tool pages. The free AI mastering page is now in Spanish too, next to German, French and Bulgarian.

Read the vintage sound guide if you want to choose a medium for your song.

Echoteka now speaks Bulgarian

Echoteka is now available in Bulgarian. Open the site at /bg to see the tool pages in Bulgarian, and the language switcher in the app header to choose it.

  • Tool pages in Bulgarian: free AI mastering, vocal remover, stem separation and audio cutter.
  • The app in Bulgarian: the mastering screen, the other tools and the studio. Where a tool text is not translated yet, it stays in English.
  • Legal page: the legal notice is in Bulgarian. The terms, privacy, refund and cookie policies stay in English.

Nothing else changed: the same tools, the same limits and the same prices.

Echoteka update: track videos in the lab, louder masters for loud tracks, German and French pages

Several changes went live today. All of them are in the lab now.

  • Track video inside the lab. Track video builds a short vertical video for Reels, Shorts and TikTok that moves to the beat of your track and cuts the hook for you. The video is built in your browser, so your file is not uploaded for it. A build costs 1 credit and needs a sign-in; rebuilding the same track within 24 hours is free.
  • Loud tracks stay loud. If your upload is already loud and you did not pick a loudness target yourself, the master is no longer turned down to −14 LUFS. It keeps the loudness close to your track, within the limit of the AI advice. Picking a target by hand still works exactly as before.
  • German and French pages. KI-Mastering, Gesang entfernen, Mastering IA and Supprimer la voix, with a language line on every page that has a translation.
  • An About page and a referral link. About Echoteka says who runs the service. Signed-in users find their invite link at /app/referral.

The full notes are in the changelog.

Echoteka 1.3: a second master, an AI label check and an honest cleanup result

Echoteka 1.3 is out. Three things changed this week, and all of them are in the lab now.

  • A second master that keeps the dynamics. If loudness has flattened your track, a slightly quieter version appears under the result. Read how it works.
  • An AI label check. Track check has an AI label mode. It reads what the file says about its generator and gives a checklist with sources for Spotify, Deezer, DistroKid and Article 50 of the EU AI Act. It does not decide whether your track is AI-made.
  • An honest cleanup result. If AI cleanup finds nothing to remove, the lab says so, your token comes back, and the daily limit is not used.

The full notes are in the changelog.

Echoteka 1.2: hear it first, master from your AI assistant

Echoteka 1.2 is out. What's new:

  • Hear it before you upload — a before/after player on a real Suno song, loudness-matched, on the home page and Suno mastering. The LUFS checker has a 20-second sample.
  • Master from Claude, Cursor or Codex — our MCP server analyzes, cleans and masters from a link, with your free daily master.
  • Clean AI artifacts and master in one run — one free switch; the cleanup downloads as MP3 or FLAC too.
  • Audio restoration for cassette, vinyl, phone and interview recordings.
  • Loudness follows your platform, cleaner low 808 notes, and AAC, AIFF and WebM uploads.
  • Invite a friend: +3 masters each, and an AI consultant on the site and in @EchotekaAIBot.

Read the release post or the full changelog.

Echoteka 1.1: a free master every day

Echoteka 1.1 is out. The headline: 1 full master every day for free, with or without an account. Signing up adds 3 bonus masters.

Also new: a free LUFS checker and mastering calculator, the Suno loudness report with data from 248 Suno exports, livelier masters at streaming loudness, breath control for voice and gentler AI artifact cleanup.

Read more in the release post or the full changelog.

Echoteka is live

Echoteka is open to everyone. It is an online mastering lab for tracks made with Suno and other AI generators, and for home recordings.

Upload a WAV or MP3 and get a release-ready master at streaming loudness, with a before/after report that measures loudness, true peak, loudness range, crest factor and stereo. AI artifact cleanup takes care of the metallic ringing and watery highs typical of AI tracks.