🚧🚧 Under (re)construction 🚧🚧

After writing the initial version of LibreTakt for a university project, I've decided to rewrite it for a slightly different use case. The new version will be structured as follows:

• a single synthesis/sampling engine running locally on a PC/laptop
• multiple remote controllers running as a web app, based on WebAssembly and egui, connected via local network

For the previous final stage of the project, check the 1.0.0 tag.

Libretakt

An open-source standalone software sampler, allowing for collaborative music creation over the network & streaming your performance via an internet radio.

Workflow and project name inspired by Elektron's Digitakt.

You can listen to a demo here

Features

• 4 voice polyphony
• Elektron-style parameter locking
• Every voice gets a separate:
  • Animated cat
  • Amp ADSR
  • Filter ADSR
  • Delay effect with adjustable send, feedback & length
  • Reverb effect with adjustable send, size & early mix
  • [TODO] an LFO
• [TODO] a master compressor

Running

1. cargo run - default sampler UI with audio engine, no synchronisation server.
2. LIBRETAKT_SERVER=ws://70.34.252 .191:8081/3f33ef73-4104-4c84-a826-11336ee24d65 cargo run --release --features enable_synchronisation - same as above, synchronisation enabled. Remember to start the server first.
3. cargo run -p server - starts the synchronisation server.
4. cargo run --example headless_ffmpeg_client | ffmpeg -f f32le -i pipe: -f mp3 - | ffmpeg -re -f mp3 -i pipe: -c copy -f flv rtmp://baczek.me/live/livestream - Headless streaming client. Again, remember to start the server first
5. For windows $env:LIBRETAKT_SERVER = "ws://70.34.252.191:8081/3f33ef73-4104-4c84-a826-11336ee24d65"
6. cp project.json C:\\Users\\mateu\\AppData\\Local\\Temp\\project.json
7. http://70.34.252.191:8080/players/srs_player.html?autostart=true&stream=livestream.flv&port=8080&schema=http

---

Note: below are the notes collected during development and some diagrams we made for the university course. Most of it is messy, some notes are in polish :)

Learning resources used during development

• Audio playback - Rodio
• UI - MacroQuad
• Music genre classification:
  • python tools overview
  • GTZAN dataset
• On mutexes and audio processing - [using locks in real-time audio processing](Using locks in real-time audio processing, safely)

Component diagram

flowchart TD
    subgraph Client
        cSeq[Sequencer]--send playback events-->cEng
        cEng[DSP engine]
        cCont[Controller]--apply state change-->cSeq
        cUI[UI]--read state-->cSeq
        cUI--send state change<br>join/create session-->cCont
    end

    cCont--send state change<br>join/create session-->sync
    sync--synchronise state<br>with other uses-->cCont

    subgraph Web-hosted
        direction RL;
        subgraph hC[Headless client - spawned per session]
            Controller--apply state change-->Sequencer
            Sequencer--send playback events-->e[DSP engine]
        end
        sync[Synchronisation<br>service]
        sync--send state change-->Controller
        e-->stream
        stream[Streaming service]
    end

Loading

Class diagram

classDiagram
    class Engine {
        sequencer: SmartPointer~Sequencer~
        voices: Vec[Voice]

        next() Rodio requirement
    }

    Engine --> Sequencer: Engine calls tick() in the sequencer, <br>triggering events and passing its voices

    class Sequencer {
        tracks: Vec[Track]
        tick(&mut voices)
    }

    class UI {
        sequencer: SmartPointer~Sequencer~
        controller: SmartPointer~Controller~
        draw()
    }

    class Controller {
        sequencer: SmartPointer~Sequencer~
        websocket_worker: Thread? idk
        set_parameter(track, pattern, step_num, step)
    }

    Controller -->Sequencer: Controller mutates the sequencer state,<br>either because of user input or because<br>of received websocket events
    Controller <-- UI: Handle parameter change events
    UI --> Sequencer: UI reads the sequencer state

    class Track {
        playback_parameters: PlaybackParameters
        patterns: Vec[Pattern],
        current_pattern: int,
        current_step: int,
    }

    Track *-- Pattern : Patterns refer to different melodies, that <br>can be dynamically switched by users

    class Pattern {
        steps: Vec[Step]
    }

    Pattern *-- Step: Step plays a sound, optionally<br> overriding playback parameters

    class PlaybackParameters {
        parameters: [u8]

        merge(Step) PlaybackParameters
    }

    Track *-- PlaybackParameters

    class Step {
        parameters: [Option~u8~]
    }

    Sequencer *-- Track

    class Voice {
        sample_provider: SampleProvider
        play_position: float
        playback_parameters: PlaybackParameters
    }

    Voice *-- PlaybackParameters
    Engine *-- Voice

    SynchronisationService <--> Controller: Synchronise changes across users<br>within the same session

    class SynchronisationService {
        sessions: Map~session_token: String, connected_clients: WebsocketConnections~
        join_or_create_session(token, nickname)
        set_parameter(track, pattern, step_num, step)
        parameter_changes_subscription(): Stream of updates(track, pattern, step_num, step)
    }

Loading

Gantt

gantt
    title Rozkład jazduni
    dateFormat  YYYY-MM-DD

    section Web
    POC serwer websocket    :w1, 2022-10-15, 7d
    POC serializacja&deserializacja Cap'n'Proto  : 2022-10-15, 7d
    Testy synchronizacji stanu za pomocą Cap'n'Proto  : 7d

    section Logika biznesowa
    Mechanizm synchronizacji stanu :crit, 2022-10-15, 7d
    Zapisywanie stanu sequencera: 7d
    Blockowanie ścieżek na jednego użytkownika: 7d

    section UI
    Widgety z wizualizacją parametrów   :crit, 2022-10-15, 7d
    Podpięcie sequencera do widgetów   :crit, 7d
    Podgląd i edycja sekwencji : 7d
    Obsługa parameter locks : 7d
    Obsługa przełączania ścieżek i patternów : 7d

    section DSP&Sequencing
    Filtry na audio :2022-10-15, 7d
    Efekty reverb + delay : 7d
    (AMP + Filter) ADSR: 7d

Loading

Do rozpisania

1. Wymagania funkcjonalne i niefunkcjnalne:

• Zdefiniowanie czemu dana technologia została wybrana ("najlepiej spełnia wymagania XXX ponieważ YYY")

2. Zamodelowanie procesów oddziaływania użytkownika z systemem
3. Opis komunikacji między komponentami w systemie
4. Rozmieszczenie komponentw systemu

Gotowy plan raportu z wykładu xd

Wstęp

1. Opis rzeczywistości, w której funkcjonuje system
2. Opis klas obiektów występujących w rzeczywistości (modelowanie)
3. Opis atrybutów tychże klas
4. Opis relacji

Opis zadań

1. Diagram przypadków użycia
2. Flow chart
3. Sequence diagram

Opis procesów

"Proces biznesowy" - opisanie jakie procesy realizuje aplikacja z perspektywy użytkownika

Opis wymiany komunikatów

Mówił o tym tak, jakby było dla niego ważne.

1. Opis/definicja wiadomości
2. Opis sekwencji wysyłania wiadomości

Opis rozmieszczenia komponentów

VPS/VM/Dockerki, inne takie

Old diagram

graph TD
    SEQ[Sequencer]
    SOUND[Sound Engine]
    UI[User interface] -->|Read state| SEQ
    UI-->|Edit state| SEQ
    MQ[Message queue]
    SEQ-->|Emit trigger| SOUND
    MQ-->|Propagate state changes from other users|SEQ
    UI-->|Send edit state event|MQ
    SOUND-->|sample state visualisation|UI

Loading