VoiceForge

Create custom voices through mixing, cloning, generation, breeding, evolution, and mutation.

Overview

VoiceForge is ORIGIN Neural's voice creation suite. It provides six methods for creating custom voices, from simple mixing of existing voices to advanced genetic breeding algorithms. All created voices are stored in your account and can be used with any synthesis endpoint.

Voice Mixing

Blend two or more existing voices together with configurable weights. The result is a new voice embedding that combines characteristics of the source voices.

POST/v1/forge/mixAUTH

Mix multiple voices with weighted blending.

Parameters

Name	Type	Description
`voices`*	array	Array of {voice_id, weight} objects. Weights should sum to 1.0.
`name`*	string	Name for the new voice.

Request Body

{
  "voices": [
    { "voice_id": "voice_01", "weight": 0.6 },
    { "voice_id": "voice_02", "weight": 0.4 }
  ],
  "name": "Custom Blend"
}

Response

{
  "voice_id": "voice_mix_abc",
  "name": "Custom Blend",
  "type": "custom",
  "parents": ["voice_01", "voice_02"]
}

Voice Cloning

Clone a voice from an audio sample. Upload a WAV or MP3 file (10-60 seconds of clear speech) and the system generates a voice embedding that matches the speaker.

For best results, use clean audio with minimal background noise. 30+ seconds of varied speech gives the best embeddings.
Cloned voices require explicit consent from the voice owner. See our usage policy for details.

POST/v1/forge/cloneAUTH

Clone a voice from an audio sample.

Parameters

Name	Type	Description
`audio`*	file	Audio file (WAV/MP3, 10-60s, clear speech).
`name`*	string	Name for the cloned voice.
`language`	string	Language hint for better extraction.Default: English

Response

{
  "voice_id": "voice_clone_xyz",
  "name": "Cloned Speaker",
  "type": "custom",
  "quality_score": 0.87
}

Voice Generation

Generate a completely new synthetic voice from a text description. Describe the desired characteristics and the system creates a matching voice.

POST/v1/forge/generateAUTH

Generate a new voice from a text description.

Parameters

Name	Type	Description
`description`*	string	Natural language description of desired voice characteristics.
`name`*	string	Name for the generated voice.
`gender`	string	Preferred gender: male, female, neutral.

Request Body

{
  "description": "A warm, authoritative male voice with a slight British accent, suitable for audiobook narration",
  "name": "British Narrator",
  "gender": "male"
}

Response

{
  "voice_id": "voice_gen_abc",
  "name": "British Narrator",
  "type": "custom"
}

Voice Breeding

VoiceBreed uses genetic algorithms to evolve voices. Select two parent voices and breed them to create offspring that inherit traits from both parents with natural variation. This is the core of ORIGIN's unique voice genetics system.

POST/v1/forge/breedAUTH

Breed two parent voices to create offspring.

Parameters

Name	Type	Description
`parent_a`*	string	Voice ID of the first parent.
`parent_b`*	string	Voice ID of the second parent.
`name`*	string	Name for the offspring voice.
`mutation_rate`	number	Mutation factor (0.0 - 1.0). Higher values = more variation.Default: 0.1

Response

{
  "voice_id": "voice_breed_xyz",
  "name": "Offspring Voice",
  "type": "evolved",
  "parents": ["voice_01", "voice_02"],
  "generation": 1,
  "traits": {
    "warmth": 0.72,
    "clarity": 0.85,
    "energy": 0.61
  }
}

Voice Evolution

Iteratively evolve a voice through multiple generations. Each generation applies selection pressure based on your feedback or automatic quality metrics.

POST/v1/forge/evolveAUTH

Evolve a voice through multiple breeding generations.

Parameters

Name	Type	Description
`voice_id`*	string	Voice ID to evolve from.
`generations`	number	Number of evolution generations.Default: 3
`target_traits`	object	Target trait values to optimize toward.
`population_size`	number	Number of candidates per generation.Default: 5

Response

{
  "evolved_voice_id": "voice_evo_abc",
  "generations_run": 3,
  "trait_improvement": {
    "warmth": "+0.12",
    "clarity": "+0.08"
  }
}

Voice Mutation

Apply random mutations to an existing voice to explore variations. Useful for discovering unexpected voice characteristics.

POST/v1/forge/mutateAUTH

Apply random mutations to a voice.

Parameters

Name	Type	Description
`voice_id`*	string	Voice ID to mutate.
`name`*	string	Name for the mutated voice.
`intensity`	number	Mutation intensity (0.0 - 1.0).Default: 0.3

Response

{
  "voice_id": "voice_mut_xyz",
  "name": "Mutated Voice",
  "type": "evolved",
  "parent": "voice_01",
  "mutation_delta": 0.27
}

Voice Quality Assessment

Evaluate the quality of a voice embedding. Returns scores for intelligibility, naturalness, speaker similarity (for cloned voices), and overall quality.

POST/v1/forge/qualityAUTH

Assess voice embedding quality.

Parameters

Name	Type	Description
`voice_id`*	string	Voice ID to evaluate.
`reference_audio`	file	Reference audio for similarity scoring (cloned voices).

Response

{
  "voice_id": "voice_clone_xyz",
  "scores": {
    "intelligibility": 0.92,
    "naturalness": 0.85,
    "speaker_similarity": 0.88,
    "overall": 0.88
  }
}

← PreviousStreaming

Next →Blockchain

VoiceForge

Create custom voices through mixing, cloning, generation, breeding, evolution, and mutation.

Overview

Voice Mixing

Blend two or more existing voices together with configurable weights. The result is a new voice embedding that combines characteristics of the source voices.

POST/v1/forge/mixAUTH

Mix multiple voices with weighted blending.

Parameters

Name	Type	Description
`voices`*	array	Array of {voice_id, weight} objects. Weights should sum to 1.0.
`name`*	string	Name for the new voice.

Request Body

{
  "voices": [
    { "voice_id": "voice_01", "weight": 0.6 },
    { "voice_id": "voice_02", "weight": 0.4 }
  ],
  "name": "Custom Blend"
}

Response

{
  "voice_id": "voice_mix_abc",
  "name": "Custom Blend",
  "type": "custom",
  "parents": ["voice_01", "voice_02"]
}

Voice Cloning

Clone a voice from an audio sample. Upload a WAV or MP3 file (10-60 seconds of clear speech) and the system generates a voice embedding that matches the speaker.

For best results, use clean audio with minimal background noise. 30+ seconds of varied speech gives the best embeddings.
Cloned voices require explicit consent from the voice owner. See our usage policy for details.

POST/v1/forge/cloneAUTH

Clone a voice from an audio sample.

Parameters

Name	Type	Description
`audio`*	file	Audio file (WAV/MP3, 10-60s, clear speech).
`name`*	string	Name for the cloned voice.
`language`	string	Language hint for better extraction.Default: English

Response

{
  "voice_id": "voice_clone_xyz",
  "name": "Cloned Speaker",
  "type": "custom",
  "quality_score": 0.87
}

Voice Generation

Generate a completely new synthetic voice from a text description. Describe the desired characteristics and the system creates a matching voice.

POST/v1/forge/generateAUTH

Generate a new voice from a text description.

Parameters

Name	Type	Description
`description`*	string	Natural language description of desired voice characteristics.
`name`*	string	Name for the generated voice.
`gender`	string	Preferred gender: male, female, neutral.

Request Body

{
  "description": "A warm, authoritative male voice with a slight British accent, suitable for audiobook narration",
  "name": "British Narrator",
  "gender": "male"
}

Response

{
  "voice_id": "voice_gen_abc",
  "name": "British Narrator",
  "type": "custom"
}

Voice Breeding

POST/v1/forge/breedAUTH

Breed two parent voices to create offspring.

Parameters

Name	Type	Description
`parent_a`*	string	Voice ID of the first parent.
`parent_b`*	string	Voice ID of the second parent.
`name`*	string	Name for the offspring voice.
`mutation_rate`	number	Mutation factor (0.0 - 1.0). Higher values = more variation.Default: 0.1

Response

{
  "voice_id": "voice_breed_xyz",
  "name": "Offspring Voice",
  "type": "evolved",
  "parents": ["voice_01", "voice_02"],
  "generation": 1,
  "traits": {
    "warmth": 0.72,
    "clarity": 0.85,
    "energy": 0.61
  }
}

Voice Evolution

Iteratively evolve a voice through multiple generations. Each generation applies selection pressure based on your feedback or automatic quality metrics.

POST/v1/forge/evolveAUTH

Evolve a voice through multiple breeding generations.

Parameters

Name	Type	Description
`voice_id`*	string	Voice ID to evolve from.
`generations`	number	Number of evolution generations.Default: 3
`target_traits`	object	Target trait values to optimize toward.
`population_size`	number	Number of candidates per generation.Default: 5

Response

{
  "evolved_voice_id": "voice_evo_abc",
  "generations_run": 3,
  "trait_improvement": {
    "warmth": "+0.12",
    "clarity": "+0.08"
  }
}

Voice Mutation

Apply random mutations to an existing voice to explore variations. Useful for discovering unexpected voice characteristics.

POST/v1/forge/mutateAUTH

Apply random mutations to a voice.

Parameters

Name	Type	Description
`voice_id`*	string	Voice ID to mutate.
`name`*	string	Name for the mutated voice.
`intensity`	number	Mutation intensity (0.0 - 1.0).Default: 0.3

Response

{
  "voice_id": "voice_mut_xyz",
  "name": "Mutated Voice",
  "type": "evolved",
  "parent": "voice_01",
  "mutation_delta": 0.27
}

Voice Quality Assessment

Evaluate the quality of a voice embedding. Returns scores for intelligibility, naturalness, speaker similarity (for cloned voices), and overall quality.

POST/v1/forge/qualityAUTH

Assess voice embedding quality.

Parameters

Name	Type	Description
`voice_id`*	string	Voice ID to evaluate.
`reference_audio`	file	Reference audio for similarity scoring (cloned voices).

Response

{
  "voice_id": "voice_clone_xyz",
  "scores": {
    "intelligibility": 0.92,
    "naturalness": 0.85,
    "speaker_similarity": 0.88,
    "overall": 0.88
  }
}

← PreviousStreaming

Next →Blockchain