FluxWave is a typed, async Lavalink v4 client for Python Discord music bots. It provides the Lavalink REST and websocket client, Discord voice integration, node pooling, queueing, filters, events, plugin helpers, autoplay, persistence, and production-oriented recovery tools.
FluxWave is 0.2.2 (beta). It is usable in real bots today; being pre-1.0,
some public APIs may still change before 1.0.
- Async Lavalink v4 REST + websocket client, fully typed (
py.typed). - Library-agnostic — works with discord.py, py-cord, nextcord, or disnake;
depends only on
aiohttp. - Multi-node pooling (
NodePool/Pool) with health-aware selection, blacklist/cooldown, automatic failover and return-to-home, hot-node draining, and parallel multi-node search. - Players — full playback controls, voice-state recovery, a tagged filter stack, opt-in crossfade (smooth volume fades between tracks), and a voice watchdog for stalled-playback recovery.
- Queue — history, loop modes, shuffle/move/dedupe, capacity limits, and serialization.
- Filters — equalizer, karaoke, timescale, a 30+ effect preset library
(nightcore, 8D, slowed + reverb, lo-fi, genre EQs…), interpolation, and
set_filters(..., seek=True). - Search —
ytsearch:/ytmsearch:/URLs/playlists, default sources, LFU caching, and in-flight request coalescing. - Events — typed payloads with
fluxwave_*and Wavelink-style aliases, plus a raw-websocket event for debugging. - Plugins — LavaSrc, LavaLyrics, SponsorBlock, and custom REST routes.
- Lyrics — fetch full/plain lyrics, plus
live_lyrics()to stream synced lines as the track plays for karaoke-style now-playing messages. - Autoplay — recommendation providers with duplicate filtering.
- Persistence & observability — in-memory and crash-safe on-disk state stores, metrics, and structured tracing.
Full documentation is hosted at fluxwave.readthedocs.io
(source in docs/). Good starting points:
- Getting Started — your first bot, step by step.
- Basic Bot and Advanced Bot — runnable examples.
- API Reference · FAQ · Migration · Changelog
The docs/guide/ folder covers players, nodes & pools, search &
autoplay, queues & filters, events, plugins, persistence & observability, and
troubleshooting.
- Python
3.11or newer. - One supported Discord library, which you install yourself:
discord.py2.4+,py-cord2.4+,nextcord2.6+, ordisnake2.9+. Like mafic and lavalink.py, FluxWave does not depend on a specific one — it detects whichever is installed. aiohttp.- A Lavalink v4 server.
FluxWave needs two things: FluxWave itself, plus any one Discord library (discord.py, py-cord, nextcord, or disnake — your choice). FluxWave auto-detects whichever one is installed, so there is nothing to configure.
If you already have a Discord library installed, just install FluxWave:
python -m pip install fluxwaveIf you are starting fresh, install FluxWave and a Discord library together
(discord.py shown here — swap in py-cord, nextcord, or disnake if you
prefer):
python -m pip install fluxwave discord.pyOr pull a Discord library in automatically with an extra
(discordpy, pycord, nextcord, or disnake):
python -m pip install "fluxwave[discordpy]"That is the whole setup. Once a Discord library is importable, import fluxwave
just works.
Why isn't the Discord library bundled? FluxWave itself depends only on
aiohttp. It deliberately does not install a Discord library for you, because discord.py and py-cord share the samediscordimport name — bundling one would overwrite and break users of the other. Bringing your own library (the same approach as mafic and lavalink.py) keeps every install clash-free.
If you happen to have more than one supported library installed at once, tell
FluxWave which to use with the FLUXWAVE_DISCORD_LIBRARY environment variable
(discord, nextcord, or disnake).
For local development:
git clone https://github.com/FluxWaveProject/fluxwave.git
cd fluxwave
python -m pip install -e ".[dev,docs]"Create a Lavalink node and use FluxPlayer as the Discord voice client:
import os
import discord
from discord.ext import commands
import fluxwave
bot = commands.Bot(command_prefix="!", intents=discord.Intents.default())
@bot.event
async def setup_hook() -> None:
node = fluxwave.Node( # describe one Lavalink server
uri=os.getenv("LAVALINK_URI", "http://127.0.0.1:2333"),
password=os.getenv("LAVALINK_PASSWORD", "youshallnotpass"),
user_id=bot.user.id, # the bot's own Discord user id
identifier="main",
)
await fluxwave.Pool.connect(nodes=[node], cache_capacity=256) # register + connect the node(s)
@bot.command()
async def play(ctx: commands.Context, *, query: str) -> None:
if ctx.author.voice is None or ctx.author.voice.channel is None: # caller must be in a voice channel
await ctx.reply("Join a voice channel first.")
return
player = ctx.voice_client # existing voice connection, if any
if not isinstance(player, fluxwave.FluxPlayer):
player = await ctx.author.voice.channel.connect(cls=fluxwave.FluxPlayer) # join voice as a FluxPlayer
result = await player.enqueue(query) # search and add tracks to the queue
if result.added == 0: # nothing matched the query
await ctx.reply("No tracks found.")
return
if player.current is None: # nothing playing yet
await player.skip(force=True) # start the first queued track
await ctx.reply(result.message) # human-friendly summary of what was added
bot.run(os.environ["DISCORD_TOKEN"])See examples/basic_bot.py for a fuller command set.
See examples/advanced_bot.py for autoplay,
filters, lyrics, persistence, metrics, and watchdog usage.
Node represents one Lavalink server. It owns REST, websocket, authentication,
session resume, route planner helpers, known Lavalink players, live Discord
players, and optional node-local search caching.
Read more: Nodes and Pools, API Reference: Node API.
node = fluxwave.Node( # describe one Lavalink server
uri="http://127.0.0.1:2333",
password="youshallnotpass",
user_id=1234567890, # the bot's own Discord user id
identifier="main",
search_cache_capacity=256, # per-node search result cache size
)
await node.connect() # open REST + websocket to LavalinkNodePool is instance-based. Pool is the global facade for bot code that
prefers Wavelink-style helpers.
Read more: Nodes and Pools, API Reference: Node API.
await fluxwave.Pool.connect(nodes=[node], cache_capacity=512) # register + connect the node(s)
tracks = await fluxwave.Pool.search("never gonna give you up") # search the default sourceFor multiple nodes:
results = await fluxwave.Pool.search_all("ytsearch:lofi") # search every node in parallel
degraded = fluxwave.Pool.get_degraded_nodes() # nodes currently unhealthy
await fluxwave.Pool.drain(old_node, cooldown=300) # stop using a node, cooldown in secondsFluxPlayer is a Discord voice protocol (compatible with discord.py, py-cord,
nextcord, and disnake). It handles Discord voice updates, sends voice state to
Lavalink, owns queues, controls playback, dispatches public events, and can
persist/restore playback state.
Read more: Player Guide, API Reference: Player API.
player = await voice_channel.connect(cls=fluxwave.FluxPlayer) # join voice as a FluxPlayer
await player.enqueue("ytsearch:lofi beats") # search and add tracks to the queue
await player.skip(force=True) # advance to the next track now
await player.set_volume(80) # volume as a percentage
await player.set_filters(fluxwave.Filters().nightcore(), seek=True) # nightcore effect, re-seek so it applies nowRead more: Queues and Filters, API Reference: Queue API.
player.queue.put(track) # add a track to the end of the queue
player.queue.move(4, 0) # move the track at index 4 to the front
player.queue.dedupe() # remove duplicate tracks
next_ten = player.queue.clear_next(10) # remove and return the next 10 tracks
duration_ms = player.queue.total_duration # total queue length in milliseconds
next_track = player.queue.get(bypass_loop=True) # pop next track, ignoring loop mode
snapshot = player.queue.to_raw_data() # serialize the queue for persistenceFluxWave dispatches both FluxWave-prefixed and Wavelink-style Discord event names for easier migration:
Read more: Events, API Reference: Events.
@bot.event
async def on_fluxwave_track_start(event: fluxwave.TrackStartEvent) -> None:
print(f"Started {event.track.title} in guild {event.guild_id}")
@fluxwave.listen("track_end")
async def on_track_end(event: fluxwave.TrackEndEvent) -> None:
print(event.reason)Read more: Persistence and Observability.
store = fluxwave.MemoryStore() # in-memory state store
state = player.save_state(extra={"source": "shutdown"}) # snapshot player + queue
await store.save(player.guild.id, state) # persist keyed by guild id
restored = await store.load(player.guild.id) # load any saved snapshot
if restored is not None:
await player.restore_state(restored) # rebuild queue and playbackRead more: Persistence and Observability.
watchdog = player.start_watchdog( # monitor for stalled playback
fluxwave.WatchdogConfig(check_interval=5.0, stagnation_threshold=12.0) # poll every 5s, recover after 12s stalled
)FluxWave is not a command framework, but it is designed to make common commands simple:
await player.enqueue("song name") # search and add to the queue
await player.play_next("song name") # queue a track to play next
await player.skip(force=True) # advance to the next track now
await player.stop(force=True) # stop and clear the queue
await player.stop(clear_queue=False) # stop current track but keep queue
await player.set_filters(fluxwave.Filters().bass_boost(), seek=True) # bass boost, re-seek so it applies now
lyrics = await player.current_lyrics() # fetch lyrics for the current track
state = player.save_state() # snapshot player + queue- Never hardcode Discord bot tokens in examples or committed code.
- Read tokens/passwords from environment variables or a secret manager.
- Treat Lavalink plugin endpoints as trusted-server APIs; validate user input before passing it to custom routes.
FluxWave expects a Lavalink v4 server reachable over HTTP or HTTPS.
Common environment variables used by examples and integration tests:
export DISCORD_TOKEN="your bot token"
export LAVALINK_URI="http://127.0.0.1:2333"
export LAVALINK_HOST="127.0.0.1"
export LAVALINK_PORT="2333"
export LAVALINK_PASSWORD="youshallnotpass"
export LAVALINK_SECURE="false"python -m pip install -e ".[dev,docs]"
.venv/bin/python -m ruff check .
.venv/bin/python -m ruff format --check .
.venv/bin/python -m mypy
.venv/bin/python -m pytestRun integration tests against your own Lavalink server:
LAVALINK_HOST=127.0.0.1 \
LAVALINK_PORT=2333 \
LAVALINK_PASSWORD=youshallnotpass \
LAVALINK_SECURE=false \
.venv/bin/python -m pytest -m integration tests/test_integration_lavalink.pyBuild docs locally:
.venv/bin/python -m sphinx -b html docs docs/_build/htmlBuild a wheel/sdist locally:
.venv/bin/python -m pip install build
.venv/bin/python -m buildFluxWave is beta-quality and pre-1.0. On top of a heavy unit-test suite, it has
been validated end-to-end against live Discord and Lavalink — playback,
multi-node failover, every plugin integration, and an extended stress soak.
Public APIs may still change before 1.0.
- Bugs and feature requests: GitHub Issues
- Before opening an issue, see Troubleshooting and the FAQ.
Contributions are welcome — see CONTRIBUTING.md for the dev setup and the lint/type/test checks. Security issues: see SECURITY.md.
FluxWave is distributed under the MIT License. See LICENSE.