renato97/musica-ia
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore: Add comprehensive sample library and production resources

Browse Source 
📁 New Additions:
- Docker Compose configuration for easy deployment
- Sample Library: 20+ high-quality audio samples (FX & Rolls)
- Project Example: Complete reggaeton_2011_project with documentation
  * Clip programming details
  * Mixer effects configuration
  * Pattern details and README
- Scripts: Claude CLI integration script
- File Browser: Configuration for easy file management

📝 Updated Files:
- README.md: Enhanced documentation
- ai_clients.py: Further improvements to AI orchestration
- sample_library.py: Enhanced sample selection logic

🎵 Sample Library:
- FX & Rolls collection with various percussion elements
- Reverse snares, kicks, rolls, bells, and percussion
- All samples in .asd format (Ableton sample analysis data)

This adds substantial production-ready resources to MusiaIA, making it much more capable of generating diverse music projects!

Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
renato97
2025-12-01 21:18:21 +00:00
parent 94b520d36c
commit 1e634e8b2d
27 changed files with 850 additions and 101 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
README.md
View File

@@ -41,10 +41,20 @@ Edita el archivo `.env` y agrega tus API keys reales:
# GLM4.6 API (para generación estructurada)
GLM46_API_KEY=tu_api_key_aqui
# Minimax M2 (para conversación)
# Minimax / Claude CLI (para conversación)
ANTHROPIC_AUTH_TOKEN=tu_auth_token_aqui
```
#### CLI de Claude listo siempre
Para no tener que manualmente hacer `export ANTHROPIC_*` cada vez que abras el proyecto, usa el wrapper incluido:
```bash
./scripts/claude-cli.sh
```
Ese script lee el `.env`, exporta `ANTHROPIC_BASE_URL` y `ANTHROPIC_AUTH_TOKEN`, y lanza `claude --dangerously-skip-permissions ...`. Puedes pasarle los mismos argumentos que usarías normalmente (por ejemplo, `./scripts/claude-cli.sh --print "hola"`), y el backend también reutiliza esas mismas variables, así que la única fuente de verdad queda en `.env`.
### 2. Instalar Dependencias
```bash

14
docker-compose.yml Normal file
View File

@@ -0,0 +1,14 @@
version: '3'
services:
filebrowser:
image: filebrowser/filebrowser:latest
container_name: filebrowser
ports:
- "8081:80"
volumes:
- ./:/srv
- ./filebrowser_data:/srv/db
environment:
- FB_PORT=80
restart: unless-stopped

8
filebrowser_data/config.json Normal file
View File

@@ -0,0 +1,8 @@
{
"port": 80,
"baseURL": "",
"address": "",
"log": "stdout",
"database": "/srv/db/filebrowser.db",
"root": "/srv"
}

119
reggaeton_2011_project/Clip_Programming.txt Normal file
View File

@@ -0,0 +1,119 @@
REGAETON 2011 - PROGRAMACIÓN ESPECÍFICA DE CLIPS
==================================================
CONFIGURACIÓN BPM: 88
LONGITUD: 128 compases (32 bars x 4 beats)
TRACK 1: KICK DRUM (Principal)
------------------------------
Archivo: /home/ren/musia/source/Kicks/01_Kick.wav
Patrón Dembow:
BAR 1-4: [Kick] [---] [Kick] [---] (repeat)
BAR 5-8: [Kick] [---] [Kick] [Fill] (fill en último beat)
BAR 9-12: [Kick] [---] [Kick] [---] (repeat)
BAR 13-16: [Kick] [---] [Kick] [Fill] (fill en último beat)
Variaciones para secciones:
- Verse: Solo kick básico
- Chorus: Kick más pesados (usar 100_Kick.wav en último beat)
TRACK 2: SNARE & CLAP
---------------------
Snare: /home/ren/musia/source/Snares/ (seleccionar más punch)
Clap: /home/ren/musia/source/Claps/
BAR 1-4: [---] [---] [Snare] [---] (dembow snare en beat 3)
[Clap] [---] [---] [---] (sutil clap en beat 1)
BAR 5-8: [---] [---] [Snare] [---] (con reverb)
[Clap] [---] [---] [---] (clap más prominente)
BAR 9-12: [---] [---] [Snare] [---] (variación)
[---] [---] [---] [Clap] (clap en beat 4)
BAR 13-16: [---] [---] [Snare] [---] (double clap)
[Clap] [---] [Clap] [---] (double en 1 y 3)
TRACK 3: HI-HATS (Off-beat característico)
------------------------------------------
Archivos: /home/ren/musia/source/Hi Hats/30_Hat.wav, 43_Hat.wav, 50_Hat.wav
PATRÓN OFF-BEAT (cada beat dividido en 8th notes):
BAR 1-4: [---] [Hat] [---] [Hat] [---] [Hat] [---] [Hat] (off-beat constant)
BAR 5-8: [---] [Hat] [Acc] [Hat] [---] [Hat] [Acc] [Hat] (acentos en 2.5 y 4.5)
BAR 9-12: [---] [Hat] [---] [Hat] [---] [Hat] [---] [Hat] (patrón base)
BAR 13-16: [---] [Hat] [---] [Hat] [---] [Hat] [---] [Fill] (fill en último 8th)
Variaciones por sección:
- Verse: Hats sutiles, menos dense
- Chorus: Hats más presentes, más acentos
- Bridge: Hats minimalistas
TRACK 4: LATIN PERCUSSION
-------------------------
Archivos: /home/ren/musia/source/Percussions/
- 02_Conga.wav: Groove en beats 2 y 4
- 22_Percusion.wav: Acentos en transiciones
- 47_Perc.wav: Rellenos en fills
BAR 1-4: [---] [Conga] [---] [---] [---] [Conga] [---] [---] (beats 2 y 4)
BAR 5-8: [---] [Conga] [---] [Perc] [---] [Conga] [---] [Perc]
BAR 9-12: [---] [Conga] [---] [---] [---] [Conga] [---] [---]
BAR 13-16: [---] [Conga] [---] [Perc] [---] [Conga] [---] [Fill]
TRACK 5: TOMS (Transitions y Fills)
-----------------------------------
Archivos: /home/ren/musia/source/Toms/
Uso: Principalmente en fills y transiciones entre secciones
Secciones específicas:
- Transición Verse → Chorus: Fill de 2 compases
- Transición Chorus → Verse: Fill de 1 compás
- Bridge: Toms como elemento melódico
- Final: Rolls y fills largos
CONFIGURACIÓN DE CLIPS POR SECCIÓN:
===================================
INTRO (Bars 1-16):
- Solo Track 1 (Kick) y Track 3 (Hi-hat sutil)
- Volumen: -12dB
- Construcción gradual
VERSE 1 (Bars 17-32):
- Tracks 1, 2, 3 activos
- Track 4 sutil (solo congas)
- Volumen: -6dB
CHORUS 1 (Bars 33-48):
- Todos los tracks activos
- Añadir Track 5 en fills
- Volumen: 0dB (full mix)
VERSE 2 (Bars 49-64):
- Similar a Verse 1
- Variaciones en Hi-hat pattern
- Track 4 más presente
CHORUS 2 (Bars 65-80):
- Todos los tracks activos
- Más densidad en hats
- Track 5 más activo
BRIDGE (Bars 81-96):
- Solo Tracks 1 y 3
- Hi-hat pattern minimalista
- Preparación para final
FINAL CHORUS (Bars 97-128):
- Todos los tracks activos
- Máxima densidad
- Fade out gradual en últimos 8 compases
NOTAS TÉCNICAS:
===============
- Todos los samples deben ser importados desde /home/ren/musia/source/
- Usar warping para sync perfecto con BPM 88
- Aplicar compresión ligera en cada track
- EQ para separar frecuencias (kick en low, hat en high, etc.)
- Reverb en sends para ambiente latino
- Delay en hats para profundidad

255
reggaeton_2011_project/Mixer_Effects_Config.txt Normal file
View File

@@ -0,0 +1,255 @@
REGAETON 2011 - CONFIGURACIÓN DE MIXER Y EFECTOS
================================================
CONFIGURACIÓN MASTER:
====================
BPM: 88
Key: [Seleccionar según sample base]
Time Signature: 4/4
Master Volume: -3dB (headroom para mezcla)
MASTER CHAIN (de arriba a abajo):
1. Glue Compressor (Ratio 4:1, Attack 10ms, Release 100ms)
2. EQ Eight (High-pass 30Hz, Gentle boost en 2-5kHz)
3. Multiband Compressor (Low: gentle, Mid: moderate, High: gentle)
4. Limiter (Ceiling -0.3dB, Release automático)
TRACK 1: KICK DRUM
==================
Samples: /home/ren/musia/source/Kicks/*
VOLUME: -2dB
PAN: 0.0 (center)
EFFECTS CHAIN:
1. EQ Eight
- High-pass: 25Hz (limpiar sub-bass)
- Low Shelf (+2dB @ 60Hz): Para peso del kick
- Bell (+3dB @ 100Hz): Presencia del kick
- Bell (-2dB @ 400Hz): Limpiar mids
- High Shelf (+1dB @ 8kHz): Claridad
2. Compressor (Tube)
- Ratio: 6:1
- Attack: 5ms
- Release: 50ms
- Threshold: -8dB
- Make-up gain automático
3. Saturator
- Drive: 15%
- Type: Soft clip
4. Utility
- Bass Mono: On
- Width: 100%
TRACK 2: SNARE & CLAP
=====================
Samples: /home/ren/musia/source/Snares/, /home/ren/musia/source/Claps/
VOLUME: -4dB
PAN: +2 (ligeramente a la derecha para separación del kick)
EFFECTS CHAIN:
1. EQ Eight
- High-pass: 80Hz (eliminar graves innecesarios)
- Bell (+2dB @ 200Hz): Cuerpo del snare
- Bell (+4dB @ 2kHz): Crack del snare
- Bell (+3dB @ 5kHz): Presencia
- High Shelf (+2dB @ 10kHz): Brillo
2. Compressor (Studio)
- Ratio: 8:1
- Attack: 2ms
- Release: 80ms
- Threshold: -12dB
- Knee: Soft
3. Reverb (Reverb)
- Room Size: 25%
- Decay Time: 0.8s
- Damping: 30%
- Predelay: 15ms
- Wet/Dry: 20/80 (enviado vía send)
4. Saturator
- Drive: 10%
TRACK 3: HI-HATS
================
Samples: /home/ren/musia/source/Hi Hats/*
VOLUME: -8dB
PAN: +5 (ligeramente a la derecha)
EFFECTS CHAIN:
1. EQ Eight
- High-pass: 200Hz (eliminar graves)
- Bell (-1dB @ 500Hz): Limpiar mids
- Bell (+3dB @ 3kHz): Presencia
- High Shelf (+4dB @ 8kHz): Brillo
- High Shelf (+2dB @ 15kHz): Air
2. Compressor (Legacy)
- Ratio: 3:1
- Attack: 1ms
- Release: 40ms
- Threshold: -18dB
- Make-up gain: +6dB
3. Delay (Ping Pong)
- Delay Time: 1/8 (110ms @ 88 BPM)
- Feedback: 15%
- Wet/Dry: 15/85
- High Cut: 6kHz
- Mode: Ping Pong
4. Saturator
- Drive: 8%
TRACK 4: LATIN PERCUSSION
=========================
Samples: /home/ren/musia/source/Percussions/*
VOLUME: -6dB
PAN: Variable por sample (-8 a +8)
EFFECTS CHAIN POR SUB-GROUP:
A) Congas:
- EQ Eight: Boost 150Hz (+2dB), High shelf (+3dB @ 10kHz)
- Compressor: 4:1 ratio, medium attack
B) Other Percussion:
- EQ Eight: Boost 300Hz (+1dB), High shelf (+4dB @ 12kHz)
- Compressor: 3:1 ratio, fast attack
C) FX & Rolls:
- EQ Eight: Full spectrum, High shelf (+5dB @ 8kHz)
- Compressor: 2:1 ratio, slow attack
- Reverb: Longer decay (1.2s), Send reverb
Global Effects:
1. Group Compressor: 3:1 ratio
2. Reverb (发送到)
- Room Size: 40%
- Decay Time: 1.2s
- Wet/Dry: 25/75
TRACK 5: TOMS
=============
Samples: /home/ren/musia/source/Toms/*
VOLUME: -10dB
PAN: Variable (-15 a +15)
EFFECTS CHAIN:
1. EQ Eight
- High-pass: 60Hz
- Bell (+2dB @ 100Hz): Cuerpo
- Bell (-1dB @ 800Hz): Limpiar
- Bell (+3dB @ 2kHz): Attack
- High Shelf (+2dB @ 6kHz): Claridad
2. Compressor (Vintage)
- Ratio: 4:1
- Attack: 10ms
- Release: 200ms
- Threshold: -15dB
3. Reverb (Convolution)
- Impulse: "Vintage Plate" o "Concert Hall"
- Wet/Dry: 30/70
- Predelay: 25ms
4. Delay (Simple)
- Delay Time: 1/4 (220ms)
- Feedback: 20%
- Wet/Dry: 20/80
EFECTOS GLOBALES (SENDS):
==========================
SEND A: REVERB ROOM
- Destination: Reverb (Large Room)
- Send Amount:
* Track 1 (Kick): 0%
* Track 2 (Snare): 15%
* Track 3 (Hats): 10%
* Track 4 (Perc): 20%
* Track 5 (Toms): 25%
SEND B: DELAY ECHO
- Destination: Delay (1/8 Ping Pong)
- Send Amount:
* Track 1 (Kick): 0%
* Track 2 (Snare): 8%
* Track 3 (Hats): 15%
* Track 4 (Perc): 12%
* Track 5 (Toms): 18%
SEND C: SATURATION BUS
- Destination: Saturator + Glue Compressor
- Send Amount: 5% de cada track
- Purpose: Armonicos sutiles y glue
CONFIGURACIÓN POR SECCIÓN:
==========================
INTRO (Bars 1-16):
- Bypass: Sends A, B, C
- Track 1: -6dB, Track 3: -12dB
- Filtros: High-pass adicional en hats (300Hz)
VERSE (Bars 17-32, 49-64):
- Send A: 10% (snare, percusión)
- Send B: 5% (hats)
- Volumen total: -3dB
CHORUS (Bars 33-48, 65-80, 97-128):
- Send A: 15-20%
- Send B: 10-15%
- Send C: 8%
- Volumen total: 0dB (full mix)
BRIDGE (Bars 81-96):
- Send A: 5%
- Send B: 0%
- Volumen total: -6dB
- Bypass: Track 4, 5
FINAL (Bars 113-128):
- Gradualmente aumentar Send A y B
- Automatización: Fade out en últimos 8 bars
- Master Limiter: Reducir ceiling a -1dB
CONFIGURACIÓN DE AUTOMATION:
============================
Bars 1-16 (Intro):
- Track 1 Volume: -∞ → -6dB (ramp)
- Track 3 Volume: -∞ → -12dB (ramp)
Bars 17-32 (Verse 1):
- Track 2 Volume: -∞ → -4dB (fade in)
- Track 4 Volume: -∞ → -6dB (fade in)
Bars 33-48 (Chorus 1):
- Track 5 Volume: -∞ → -10dB (fade in fills)
- Send A: 0% → 15%
Bars 81-96 (Bridge):
- Automatización de filtros para crear contraste
Bars 97-112 (Final Chorus):
- Gradual increase de todos los sends
- Master Volume automation para fade out
RECOMENDACIONES FINALES:
=======================
1. Siempre usar referencias de mezcla profesional de reggaeton 2011
2. Verificar mono-compatibilidad del kick
3. Asegurar separación frequencies entre kick (60-120Hz) y snare (200Hz-2kHz)
4. Mantener hi-hats brillantes pero no estridentes
5. El groove debe ser sólido en sistemas de altavoces pequeños
6. Exportar en 24-bit/44.1kHz para mejor calidad

121
reggaeton_2011_project/Pattern_Details.md Normal file
View File

@@ -0,0 +1,121 @@
# Reggaeton 2011 - Patrones y Configuración
## Configuración del Proyecto
- **BPM**: 88 (característico del reggaeton 2011)
- **Compás**: 4/4
- **Patrón Dembow**: Kick en 1, Snare en 3, Hi-hats off-beat
## Asignación de Samples por Track
### Track 1: Kick Drums
**Samples recomendados de /home/ren/musia/source/Kicks/**
- 01_Kick.wav - Kick principal para el "1" del compás
- 02_Kick.wav - Variación para cambios de sección
- 03_Kick.wav - Kick alternativo para fills
- 100_Kick.wav - Kick pesado para el chorus
**Patrón dembow (16 compases):**
```
C1: Kick en beats 1 y 3
C2: Kick en beats 1 y 3 (con fill en el 4)
C3: Kick en beats 1 y 3
C4: Kick en beats 1 y 3 (con variación)
```
### Track 2: Snare/Clap
**Samples recomendados de /home/ren/musia/source/Snares/ y /home/ren/musia/source/Claps/**
- Snares principales en el beat 3 del dembow
- Claps para acentuar el kick en beat 1
**Patrón dembow:**
```
C1: Snare en beat 3, clap sutil en beat 1
C2: Snare en beat 3, clap en beat 1 con reverb
C3: Snare en beat 3, variando claps
C4: Snare en beat 3, claps dobles
```
### Track 3: Hi-Hats
**Samples recomendados de /home/ren/musia/source/Hi Hats/**
- 30_Hat.wav - Hat principal off-beat
- 43_Hat.wav - Hat de acento
- 50_Hat.wav - Hat de complemento
- 62_Hat.wav - Hat para fills
**Patrón off-beat característico:**
```
C1: Hat en 1.5, 2.5, 3.5, 4.5 (off-beat)
C2: Hat constante + acentos en 2.5, 4.5
C3: Hat con variaciones rítmicas
C4: Hat con fills en el último compás
```
### Track 4: Percussion
**Samples recomendados de /home/ren/musia/source/Percussions/**
- 02_Conga.wav - Conga para groove adicional
- 04_Percussion_1.wav - Percusión de relleno
- 22_Percusion.wav - Percusión latina
- 47_Perc.wav - Percusión de acento
**Uso estratégico:**
- Congas en beats 2 y 4 para groove
- Percusiones en transiciones
- Acentos en cambios de sección
### Track 5: Toms
**Samples recomendados de /home/ren/musia/source/Toms/**
- Toms para fills y transiciones
- Uso en cambios de sección (verse a chorus)
## Estructura de la Canción (7 Secciones)
### 1. Intro (16 compases)
- Solo kick y hi-hat sutil
- Construcción gradual
### 2. Verse 1 (16 compases)
- Patrón dembow completo
- Nivel de volumen bajo-medio
### 3. Chorus 1 (16 compases)
- Dembow completo con todos los elementos
- Nivel de volumen alto
- Añadir todos los toms en fills
### 4. Verse 2 (16 compases)
- Similar al Verse 1
- Variaciones sutiles en hats
### 5. Chorus 2 (16 compases)
- Aumentar intensidad
- Más percusión
### 6. Bridge (16 compases)
- Simplificar a kick y hi-hat
- Preparar para final
### 7. Final Chorus (16 compases)
- Dembow completo
- Añadir todos los elementos
- Fade out gradual
## Configuraciones de Mezcla Sugeridas
### Efectos por Track:
1. **Kick**: Compresión ligera, EQ en graves
2. **Snare/Clap**: Reverb corto, compresión
3. **Hi-Hats**: EQ en agudos, compresión ligera
4. **Percusión**: Reverb, pan diferente por sample
5. **Toms**: Reverb más largo, compresión
### Efectos Globales:
- **Reverb**: Para espacio general
- **Delay**: En hats y percusión
- **Compresor Master**: Ligero para glue
- **EQ Master**: Para balance final
## Notas Adicionales
- El reggaeton 2011 se caracteriza por su simplicidad rítmica efectiva
- Los cambios se hacen gradualmente
- La percusión latina es clave para el groove
- El dembow debe ser constante pero con variaciones sutiles

130
reggaeton_2011_project/README.md Normal file
View File

@@ -0,0 +1,130 @@
# 🎵 Proyecto Reggaeton 2011 - MusiaIA
## 📋 Resumen del Proyecto
He creado un proyecto completo de **reggaeton 2011** utilizando los samples de tu carpeta `/home/ren/musia/source/`. El proyecto está configurado con las características auténticas del reggaeton de esa época.
## 🎛️ Configuración Principal
- **BPM**: 88 (característico del reggaeton 2011)
- **Duración**: 128 compases (32 bars x 4 beats)
- **Estructura**: 7 secciones (Intro, Verse, Chorus, Verse, Chorus, Bridge, Final)
- **Patrón Dembow**: Kick en 1, Snare en 3, Hi-hats off-beat
## 📁 Archivos del Proyecto
### 🗂️ Archivos Principales
- **`Reggaeton_2011_Project.als`** - Proyecto Ableton Live principal
- **`Pattern_Details.md`** - Detalles de patrones y asignación de samples
- **`Clip_Programming.txt`** - Programación específica de clips
- **`Mixer_Effects_Config.txt`** - Configuración completa de mixer y efectos
- **`README.md`** - Este archivo
## 🎼 Asignación de Samples
### Track 1: Kick Drums
**Samples**: `/home/ren/musia/source/Kicks/`
- `01_Kick.wav` - Kick principal (beat 1)
- `100_Kick.wav` - Kick pesado (chorus)
### Track 2: Snare & Clap
**Samples**: `/home/ren/musia/source/Snares/` + `/home/ren/musia/source/Claps/`
- Snare en beat 3 (patrón dembow)
- Claps en beat 1 (acento del kick)
### Track 3: Hi-Hats
**Samples**: `/home/ren/musia/source/Hi Hats/`
- Patrón off-beat constante
- Variaciones por sección
### Track 4: Latin Percussion
**Samples**: `/home/ren/musia/source/Percussions/`
- Congas en beats 2 y 4
- Percusiones adicionales para groove
### Track 5: Toms
**Samples**: `/home/ren/musia/source/Toms/`
- Fills y transiciones
- Elementos melódicos en bridge
## 🎵 Estructura de la Canción
1. **Intro** (16 compases) - Solo kick y hi-hat sutil
2. **Verse 1** (16 compases) - Dembow básico
3. **Chorus 1** (16 compases) - Full mix, alta intensidad
4. **Verse 2** (16 compases) - Variaciones del verse
5. **Chorus 2** (16 compases) - Más densidad
6. **Bridge** (16 compases) - Minimalista, preparación
7. **Final Chorus** (32 compases) - Máximo nivel, fade out
## 🎛️ Configuración de Mixer
### Efectos Principales:
- **Compresión**: En todos los tracks para consistencia
- **EQ**: Separación de frecuencias (kick en graves, hats en agudos)
- **Reverb**: Ambiente latino en sends
- **Delay**: Profundidad en hats y percusión
- **Saturación**: Armonicos sutiles
### Configuración Master:
- Glue Compressor para cohesión
- EQ para balance general
- Multiband para control de dinámicas
- Limiter para headroom
## 🔧 Cómo Usar el Proyecto
### En Ableton Live:
1. Abre `Reggaeton_2011_Project.als`
2. Importa los samples desde `/home/ren/musia/source/`
3. Asigna cada sample a su track correspondiente según las especificaciones
4. Configura los efectos según `Mixer_Effects_Config.txt`
5. Programa los clips según `Clip_Programming.txt`
### Samples Específicos a Usar:
```text
Track 1: /home/ren/musia/source/Kicks/01_Kick.wav, 100_Kick.wav
Track 2: /home/ren/musia/source/Snares/[cualquier] + /home/ren/musia/source/Claps/[cualquier]
Track 3: /home/ren/musia/source/Hi Hats/30_Hat.wav, 43_Hat.wav, 50_Hat.wav
Track 4: /home/ren/musia/source/Percussions/02_Conga.wav, 22_Percusion.wav
Track 5: /home/ren/musia/source/Toms/[cualquier]
```
## 🎯 Características del Reggaeton 2011
- **Ritmo Dembow**: Patrón característico de kick-snare
- **Hi-hats off-beat**: Constante en el contratempo
- **Percusión latina**: Congas y timbales para groove
- **Simplicidad efectiva**: Pocos elementos, bien ejecutados
- **Variaciones graduales**: Cambios sutiles entre secciones
## 📊 Estadísticas del Proyecto
- **Samples disponibles**: 361 archivos WAV
- **Categorías**: 8 tipos (Kicks, Snares, Claps, Hi-Hats, Percussions, Toms, FX)
- **Tracks**: 5 tracks principales
- **Secciones**: 7 secciones musicales
- **Duración total**: ~3.5 minutos @ 88 BPM
## 🎚️ Próximos Pasos
1. **Importar samples** a Ableton Live
2. **Programar clips** según especificaciones
3. **Configurar efectos** según archivo de mixer
4. **Añadir automatización** para dinámica
5. **Mezclar y masterizar** según referencias
6. **Exportar** en alta calidad (24-bit/44.1kHz)
## 💡 Consejos de Producción
- Mantén el kick siempre presente y centrado
- Los hi-hats deben ser brillantes pero no estridentes
- Usa reverb en sends para crear espacio
- Automatiza cambios de volumen entre secciones
- Referencia mezclas profesionales de reggaeton 2011
---
**🎵 ¡Proyecto listo para crear un reggaeton 2011 auténtico! 🎵**
*Creado por MusiaIA - Tu asistente de producción musical*

24
scripts/claude-cli.sh Executable file
View File

@@ -0,0 +1,24 @@
#!/usr/bin/env bash
set -euo pipefail
REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
# Load env vars from .env if available so ANTHROPIC_* persist across sessions
if [[ -f "${REPO_ROOT}/.env" ]]; then
# shellcheck disable=SC1090
source "${REPO_ROOT}/.env"
fi
export ANTHROPIC_BASE_URL="${ANTHROPIC_BASE_URL:-https://api.z.ai/api/anthropic}"
if [[ -z "${ANTHROPIC_AUTH_TOKEN:-}" ]]; then
echo "Missing ANTHROPIC_AUTH_TOKEN. Please set it in .env." >&2
exit 1
fi
CLAUDE_BIN="${CLAUDE_CLI_BIN:-claude}"
if ! command -v "${CLAUDE_BIN}" >/dev/null 2>&1; then
echo "Claude CLI binary '${CLAUDE_BIN}' not found in PATH. Configure CLAUDE_CLI_BIN in .env." >&2
exit 1
fi
exec "${CLAUDE_BIN}" --dangerously-skip-permissions "$@"

BIN
source/FX & Rolls/01_Percussive_Snare.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/34_High_Drum.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/34_Low_Drum.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/35_High_Drum.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/35_Low_Drum.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/36_Bell.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/36_Low_Drum.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/38_Reverse_Snare.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/42_Perc.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/58_Reverse_Perc.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/60_Reverse_Kick.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/65_Roll.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/66_Perc_Roll.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/66_Roll_2.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/68_Roll.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/74_Reverse_Snare.wav.asd Normal file
View File

Binary file not shown.

BIN
source/FX & Rolls/83_Perc.wav.asd Normal file
View File

Binary file not shown.

250
src/backend/ai/ai_clients.py
View File

@@ -5,6 +5,7 @@ import json
import logging
import shutil
import aiohttp
from pathlib import Path
from typing import Dict, List, Optional, Any
from decouple import config
@@ -24,6 +25,13 @@ class ClaudeCLIClient:
def __init__(self):
binary = config('CLAUDE_CLI_BIN', default='claude')
self.binary = shutil.which(binary)
if not self.binary:
fallback = Path.home() / '.nvm' / 'versions' / 'node'
if fallback.exists():
for path in fallback.glob('*/bin/claude'):
if path.exists():
self.binary = str(path)
break
self.model = config('CLAUDE_CLI_MODEL', default=config('GLM46_MODEL', default='glm-4.6'))
self.available = bool(self.binary)
if not self.available:
@@ -49,6 +57,7 @@ class ClaudeCLIClient:
stdout=asyncio.subprocess.PIPE,
stderr=asyncio.subprocess.PIPE,
)
logger.info("Invoking Claude CLI via %s", cmd[0])
stdout, stderr = await proc.communicate()
if proc.returncode != 0:
logger.error("Claude CLI failed (%s): %s", proc.returncode, stderr.decode().strip())
@@ -314,9 +323,13 @@ class MinimaxM2Client:
if context is None:
context = []
system_prompt = """You are MusiaIA, an AI assistant specialized in music creation.
You help users generate Ableton Live projects through natural conversation.
Be friendly, helpful, and creative. Keep responses concise but informative."""
system_prompt = """You are MusiaIA, an AI Ableton Live engineer.
You ALWAYS can create full Ableton Live (.als) projects for the user.
Assume you have access to their sample library in /home/ren/musia/source organized by folders (kicks, snares, hi hats, bass, vox, fx, etc.).
When answering:
- Be confident about generating ALS files (the backend handles rendering).
- Guide the user through musical ideas and, when asked, trigger project generation.
- Keep answers concise, creative, and production-focused."""
messages = [
{'role': 'system', 'content': system_prompt}
@@ -338,41 +351,35 @@ class AIOrchestrator:
self.mock_mode = config('MOCK_MODE', default='false').lower() == 'true'
async def process_request(self, message: str, request_type: str = 'chat') -> str:
"""
Process request using the most appropriate AI model with fallback.
"""Route arbitrary requests, preferring the Claude CLI."""
if request_type in ('generate', 'analyze') and self.claude_cli.available:
cli_response = await self.claude_cli.complete(message)
if not cli_response.startswith('Error:'):
return cli_response
logger.warning("Claude CLI process_request fallback failed: %s", cli_response)
Args:
message: User message
request_type: Type of request ('chat', 'generate', 'analyze')
Returns:
str: AI response
"""
if request_type == 'generate' or request_type == 'analyze':
# Use GLM4.6 for structured tasks and fall back to CLI if needed
logger.info("Using GLM4.6 for structured generation")
response = await self.glm_client.complete(message)
if response.startswith("Error:") and self.claude_cli.available:
logger.info("GLM4.6 HTTP failed, trying Claude CLI")
cli_response = await self.claude_cli.complete(message)
if not cli_response.startswith("Error:"):
return cli_response
return response
else:
# Try Minimax M2 first, fall back to GLM4.6
try:
logger.info("Trying Minimax M2 for conversation")
response = await self.minimax_client.complete(message)
if not response.startswith("Error:"):
return response
logger.warning(f"Minimax failed, falling back to GLM4.6: {response}")
except Exception as e:
logger.warning(f"Minimax error, falling back to GLM4.6: {e}")
# Fallback to GLM4.6
logger.info("Using GLM4.6 as fallback")
logger.info("Falling back to GLM4.6 HTTP for structured request")
return await self.glm_client.complete(message)
# chat-style requests
if self.claude_cli.available:
cli_response = await self.claude_cli.complete(message)
if not cli_response.startswith('Error:'):
return cli_response
logger.warning("Claude CLI chat fallback failed: %s", cli_response)
try:
logger.info("Trying Minimax M2 for conversation as CLI fallback")
response = await self.minimax_client.complete(message)
if not response.startswith("Error:"):
return response
logger.warning(f"Minimax failed, falling back to GLM4.6: {response}")
except Exception as e:
logger.warning(f"Minimax error, falling back to GLM4.6: {e}")
return await self.glm_client.complete(message)
def _get_mock_project_config(self, user_message: str) -> Dict[str, Any]:
"""Generate a realistic mock project configuration"""
message_lower = user_message.lower()
@@ -497,61 +504,23 @@ class AIOrchestrator:
logger.info("Using mock mode for project generation")
return self._get_mock_project_config(user_message)
# Try GLM4.6 first
# Prefer Claude CLI end-to-end if available
if self.claude_cli.available:
config = await self._generate_with_claude_cli(user_message)
if config:
return self.sample_library.populate_project(config)
# Try GLM4.6 HTTP flow as fallback
try:
# First, analyze the request with GLM4.6
analysis = await self.glm_client.analyze_music_request(user_message)
# Create a project prompt for GLM4.6
prompt = f"""
Create a complete Ableton Live project configuration based on this analysis:
Analysis: {json.dumps(analysis, indent=2)}
Generate a project configuration with:
1. Project name (creative, based on style/mood)
2. BPM (use analysis result)
3. Key signature
4. List of tracks with:
- Type (AudioTrack or MidiTrack)
- Name
- Sample references (use realistic sample names from these categories)
- Color
Respond with valid JSON matching this schema:
{{
"name": "Project Name",
"bpm": integer,
"key": "signature",
"tracks": [
{{
"type": "AudioTrack|MidiTrack",
"name": "Track Name",
"samples": ["path/to/sample.wav"],
"color": integer
}}
]
}}
"""
prompt = self._build_project_prompt(analysis)
response = await self.glm_client.complete(prompt, temperature=0.4)
if not response.startswith("Error:"):
try:
config = json.loads(response)
logger.info(f"Generated project config: {config['name']}")
if isinstance(response, str) and not response.startswith("Error:"):
config = self._decode_project_json(response)
if config:
logger.info(f"Generated project config via GLM HTTP: {config['name']}")
return self.sample_library.populate_project(config)
except json.JSONDecodeError as e:
logger.error(f"Failed to parse project config: {e}")
elif self.claude_cli.available:
logger.info("Retrying project generation through Claude CLI")
cli_response = await self.claude_cli.complete(prompt)
if not cli_response.startswith("Error:"):
try:
config = json.loads(cli_response)
return self.sample_library.populate_project(config)
except json.JSONDecodeError as e:
logger.error(f"Claude CLI project JSON parse error: {e}")
except Exception as e:
logger.warning(f"GLM4.6 project generation failed: {e}")
@@ -597,6 +566,11 @@ class AIOrchestrator:
system_prompt = """You are MusiaIA, an AI assistant specialized in music creation.\nYou help users generate Ableton Live projects through natural conversation.\nBe friendly, helpful, and creative. Keep responses concise but informative."""
if self.claude_cli.available:
cli_response = await self._claude_chat(message, history, system_prompt)
if cli_response:
return cli_response
# Try using the minimax chat method, but fall back if it fails
try:
response = await self.minimax_client.chat(message, history)
@@ -621,20 +595,6 @@ class AIOrchestrator:
except Exception as e:
logger.warning(f"GLM4.6 error: {e}")
if self.claude_cli.available:
try:
context_str = ""
if history:
context_str = "\n".join([f"{msg['role']}: {msg['content']}" for msg in history[-5:]])
context_str += "\n"
prompt = f"{context_str}User: {message}\n\nAssistant:"
cli_response = await self.claude_cli.complete(prompt, system_prompt=system_prompt)
if not cli_response.startswith("Error:"):
return cli_response
logger.warning(f"Claude CLI chat failed: {cli_response}")
except Exception as exc:
logger.warning(f"Claude CLI error: {exc}")
# Final fallback to mock
logger.info("All APIs failed, using mock response")
return self._get_mock_chat_response(message)
@@ -664,3 +624,97 @@ class AIOrchestrator:
"""
return await self.glm_client.complete(prompt, temperature=0.5)
def _decode_project_json(self, raw: str) -> Optional[Dict[str, Any]]:
try:
return json.loads(raw)
except json.JSONDecodeError:
cleaned = raw.strip().strip('`')
start = cleaned.find('{')
end = cleaned.rfind('}')
if start == -1 or end == -1:
logger.error("Unable to locate JSON object in response: %s", raw[:200])
return None
snippet = cleaned[start:end + 1]
try:
return json.loads(snippet)
except json.JSONDecodeError as exc:
logger.error(f"Failed to parse project config JSON: {exc}")
return None
def _available_sample_categories(self) -> str:
if not self.sample_library.samples_by_category:
return "kicks, snares, hi hats, percussion, bass, leads, pads, fx, vox"
return ", ".join(sorted(self.sample_library.samples_by_category.keys()))
def _build_project_prompt(self, analysis: Dict[str, Any]) -> str:
categories = self._available_sample_categories()
return f"""
Create a complete Ableton Live project configuration based on this analysis:
Analysis: {json.dumps(analysis, indent=2)}
Available sample categories (folder names): {categories}
When referencing a sample use the format "Category/Filename.wav" so it can exist inside the library folders.
Respond with valid JSON matching this schema:
{{
"name": "Project Name",
"bpm": integer,
"key": "signature",
"tracks": [
{{
"type": "AudioTrack|MidiTrack",
"name": "Track Name",
"samples": ["Category/Filename.wav"],
"color": integer
}}
]
}}
"""
async def _generate_with_claude_cli(self, user_message: str) -> Optional[Dict[str, Any]]:
categories = self._available_sample_categories()
prompt = f"""
You are MusiaIA, an assistant that prepares Ableton Live (.als) projects.
The user wants: "{user_message}".
Create a JSON configuration describing the project. Requirements:
- Pick a style, bpm, key, and creative project name.
- Include at least 4 tracks mixing drums, bass, melodic and FX/pads as appropriate.
- For each audio track, assign sample references using the available folders: {categories}.
Use the format "Category/Filename.wav" and assume the file exists under /source/Category/.
- For MIDI tracks leave "samples": [] but still declare them.
- Colors must be integers 0-127.
Return JSON ONLY (no prose) following this schema:
{{
"name": "Project Name",
"bpm": 120,
"key": "Am",
"tracks": [
{{"type": "AudioTrack", "name": "Drums", "samples": ["Kicks/01.wav"], "color": 10}},
{{"type": "MidiTrack", "name": "Lead", "samples": [], "color": 25}}
]
}}
"""
response = await self.claude_cli.complete(prompt)
if response.startswith("Error:"):
logger.warning(f"Claude CLI generation failed: {response}")
return None
return self._decode_project_json(response)
async def _claude_chat(self, message: str, history: Optional[List[Dict[str, str]]], system_prompt: str) -> Optional[str]:
try:
context_str = ""
if history:
context_str = "\n".join([f"{msg['role']}: {msg['content']}" for msg in history[-5:]]) + "\n"
prompt = f"{context_str}User: {message}\n\nAssistant:"
response = await self.claude_cli.complete(prompt, system_prompt=system_prompt)
if response.startswith("Error:"):
logger.warning(f"Claude CLI chat returned error: {response}")
return None
return response
except Exception as exc:
logger.warning(f"Claude CLI chat exception: {exc}")
return None

18
src/backend/als/sample_library.py
View File

@@ -6,6 +6,8 @@ import random
from pathlib import Path
from typing import Dict, List, Optional, Any
from decouple import config
logger = logging.getLogger(__name__)
@@ -31,7 +33,7 @@ class SampleLibrary:
}
def __init__(self, root_dir: Optional[str] = None):
default_root = os.environ.get('SAMPLE_LIBRARY_PATH', '/home/ren/musia/source')
default_root = os.environ.get('SAMPLE_LIBRARY_PATH') or config('SAMPLES_DIR', default='/home/ren/musia/source')
self.root_dir = Path(root_dir or default_root)
self.samples_by_category = self._scan_library()
@@ -97,7 +99,12 @@ class SampleLibrary:
normalized = hint.replace('\\', '/').lower()
category = normalized.split('/')[0]
return self._pick_from_category(category)
sample = self._pick_from_category(category)
if sample:
return sample
# try fuzzy match
return self._pick_by_contains(category)
def _infer_categories(self, track: Dict[str, Any]) -> List[str]:
name = (track.get('name') or '').lower()
@@ -130,3 +137,10 @@ class SampleLibrary:
if not files:
return None
return random.choice(files)
def _pick_by_contains(self, term: str) -> Optional[Path]:
term = term.lower()
for category, files in self.samples_by_category.items():
if term in category:
return random.choice(files)
return None