Voice Profiles & Cloning

A voice profile is the central object in the Voices tab of PeachDesk Voice Studio. It captures everything Voice Studio needs to speak as a particular voice: the engine/model, the language, optional reference audio for cloning, an avatar, a default effects chain, and an optional personality. There are three kinds of profile:

Clone from audio (voice_type: "cloned") — a custom voice built from one or more reference samples you upload.
Built-in voice (voice_type: "preset") — a hosted voice selected from a model’s preset catalog. vaani-indic (VAANI-INDIC) is the one served engine that ships built-in speakers. The model runs on PeachDesk’s GPU infrastructure; you simply pick it from a dropdown.
Designed voice (voice_type: "designed") — a voice generated from a text design_prompt.

All endpoints below are exposed through the PeachDesk API under the base path /api/voice. For example, POST /profiles is called as POST /api/voice/profiles. Every request must include your PeachDesk API key in the X-API-Key header (an interactive session token in Authorization: Bearer also works — see Authentication):

X-API-Key: bibha_your_api_key

Object: Voice Profile

The VoiceProfileResponse shape is returned by most endpoints on this page.

string

Unique profile identifier.

string

Display name shown in the Voices tab.

string | null

Optional free-text description.

string

ISO language code (e.g. en, es, ja).

string | null

Internal path to the avatar image, if one has been uploaded. Fetch it via the avatar endpoint.

EffectConfig[] | null

Default effects chain applied to takes from this voice, or null if none.

string

One of cloned, preset, or designed.

string | null

For built-in voices, the engine the preset belongs to.

string | null

For built-in voices, the selected preset voice id.

string | null

For designed voices, the prompt used to generate the voice.

string | null

Default generation engine for this profile.

string | null

Optional personality description used by the compose feature.

integer

Number of generations produced with this profile.

integer

Number of reference samples attached to this profile.

string (date-time)

Creation timestamp.

string (date-time)

Last update timestamp.

An EffectConfig is { "type": string, "enabled": boolean, "params": object }.

Managing profiles

Create a profile

POST /api/voice/profiles Create a new voice profile. The profile starts empty; for a Clone from audio voice you then add one or more reference samples (see below).

string

required

Display name (1–100 characters).

string

Optional description (max 500 characters).

string

default:"en"

Language code. Defaults to en.

string

default:"cloned"

One of cloned, preset, or designed.

string

Engine name when voice_type is preset (max 50 chars).

string

Built-in voice id when voice_type is preset (max 100 chars).

string

Prompt describing the voice when voice_type is designed (max 2000 chars).

string

Default generation engine for this profile (max 50 chars).

string

Optional personality used by the compose feature (max 2000 chars).

Returns a Voice Profile object.

curl -X POST https://app.peachdeskai.com/api/voice/profiles \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Support Agent",
    "description": "Friendly support voice",
    "language": "en",
    "voice_type": "cloned"
  }'

curl -X POST https://app.peachdeskai.com/api/voice/profiles \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Narrator",
    "language": "hi",
    "voice_type": "preset",
    "preset_engine": "vaani-indic",
    "preset_voice_id": "hi_female"
  }'

List profiles

GET /api/voice/profiles List all voice profiles for your workspace. Returns an array of Voice Profile objects.

curl https://app.peachdeskai.com/api/voice/profiles \
  -H "X-API-Key: $PEACHDESK_API_KEY"

Get a profile

GET /api/voice/profiles/ Fetch a single profile by id.

string

required

The profile id.

Returns a Voice Profile object, or 404 if not found.

Update a profile

PUT /api/voice/profiles/ Update a profile’s fields. The body is the same shape as Create a profile.

string

required

The profile id.

Returns the updated Voice Profile object.

curl -X PUT https://app.peachdeskai.com/api/voice/profiles/prof_123 \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "name": "Support Agent (US)", "language": "en", "voice_type": "cloned" }'

Delete a profile

DELETE /api/voice/profiles/ Permanently delete a profile and its samples.

string

required

The profile id.

Response

{ "message": "Profile deleted successfully" }

Built-in voices

Built-in voices are hosted models that run on PeachDesk’s GPU infrastructure. You don’t install or download anything — pick an engine and voice from the dropdown in the Voices tab, or query the catalog below.

List built-in voices for an engine

GET /api/voice/profiles/presets/ Return the catalog of preset voices for a given engine. vaani-indic (VAANI-INDIC) is the only engine with built-in speakers — 16 of them, a female and a male for each of eight Indian languages. The other served engines clone from a reference clip and ship no speaker table, so they return an empty list, as does an unknown engine.

string

required

Engine name. Use vaani-indic for the built-in catalog.

string

The engine you queried.

object[]

Array of voices, each { "voice_id", "name", "gender", "language", "description" }. description is a short use-case line, or null where none is curated.

curl https://app.peachdeskai.com/api/voice/profiles/presets/vaani-indic \
  -H "X-API-Key: $PEACHDESK_API_KEY"

Response

{
  "engine": "vaani-indic",
  "voices": [
    {
      "voice_id": "hi_female",
      "name": "Hindi Female",
      "gender": "female",
      "language": "hi",
      "description": "Warm conversational Hindi female — support calls and everyday narration"
    },
    {
      "voice_id": "hi_male",
      "name": "Hindi Male",
      "gender": "male",
      "language": "hi",
      "description": "Steady Hindi male — announcements, IVR prompts and explainers"
    }
  ]
}

Use a voice_id from this list as the preset_voice_id when creating a built-in voice profile.

List the curated built-in catalog

GET /api/voice/profiles/presets/curated Return the same built-in voices grouped for a language-first picker: a voices array plus a languages index carrying the count per language code, sorted by descending count then code.

object[]

Every curated voice, each { "name", "engine", "voice_id", "gender", "language", "persona" }.

object[]

Per-language counts, each { "language", "count" }.

curl https://app.peachdeskai.com/api/voice/profiles/presets/curated \
  -H "X-API-Key: $PEACHDESK_API_KEY"

Preview a built-in voice

GET /api/voice/profiles/presets///preview Serve a short pre-rendered sample WAV for a built-in voice, in that voice’s own language. The first request renders and caches the sample; later ones are served straight from cache.

string

required

Engine name, e.g. vaani-indic.

string

required

Built-in voice id, e.g. ta_male.

Response: audio/wav. Returns 404 if the voice is unknown or has no preview, and 503 with a Retry-After header if the render is temporarily unavailable.

Reference samples (cloning)

Reference samples are the short clips of audio that a Clone from audio profile learns from. Each sample pairs an audio file with the exact text spoken in it.

Add a sample

POST /api/voice/profiles//samples Upload a reference clip and its transcript. The request is multipart/form-data.

string

required

The profile to attach the sample to.

file

required

Audio file. Accepted formats: wav, mp3, m4a, ogg, flac, aac, webm, opus. Max 50 MB.

string

required

Exact transcript of the clip (1–1000 characters).

Returns a ProfileSampleResponse:

string

Sample id.

string

Parent profile id.

string

Internal storage path of the processed audio.

string

The transcript you provided.

curl -X POST https://app.peachdeskai.com/api/voice/profiles/prof_123/samples \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -F "file=@sample.wav" \
  -F "reference_text=Thanks for calling PeachDesk, how can I help you today?"

Files larger than 50 MB are rejected with 413 Payload Too Large. Keep reference clips short and clean — a few seconds of clear speech works best.

List samples

GET /api/voice/profiles//samples Return all samples attached to a profile.

string

required

The profile id.

Returns an array of ProfileSampleResponse objects.

Update a sample’s transcript

PUT /api/voice/profiles/samples/ Correct the reference_text of an existing sample.

string

required

The sample id.

string

required

New transcript (1–1000 characters).

Returns the updated ProfileSampleResponse, or 404 if the sample is not found.

Delete a sample

DELETE /api/voice/profiles/samples/ Remove a single sample.

string

required

The sample id.

Response

{ "message": "Sample deleted successfully" }

Avatars

Each voice can have an avatar image, shown beside the voice in the Voices tab.

Upload an avatar

POST /api/voice/profiles//avatar Upload or replace the avatar image. Request is multipart/form-data.

string

required

The profile id.

file

required

Image file.

Returns the updated Voice Profile object (with avatar_path set).

curl -X POST https://app.peachdeskai.com/api/voice/profiles/prof_123/avatar \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -F "file=@avatar.png"

Get an avatar

GET /api/voice/profiles//avatar Return the avatar image binary. Responds 404 if the profile has no avatar.

string

required

The profile id.

Delete an avatar

DELETE /api/voice/profiles//avatar Remove the avatar from a profile.

string

required

The profile id.

Response

{ "message": "Avatar deleted successfully" }

Default effects

Voice Studio applies effects as versioned takes. A profile can carry a default effects chain so every generation from that voice starts with the same processing applied.

Set or clear the default effects chain

PUT /api/voice/profiles//effects Set the default effects chain, or pass null to clear it. The chain is validated before it is saved.

string

required

The profile id.

EffectConfig[] | null

required

An ordered list of effects, or null to remove the default chain. Each entry is { "type": string, "enabled": boolean, "params": object }.

Returns the updated Voice Profile object. An invalid chain returns 400.

curl -X PUT https://app.peachdeskai.com/api/voice/profiles/prof_123/effects \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "effects_chain": [
      { "type": "reverb", "enabled": true, "params": { "room_size": 0.4 } },
      { "type": "eq", "enabled": true, "params": { "low": 2, "high": -1 } }
    ]
  }'

Channel assignment

Profiles can be assigned to channels so the right voice is used in the right context.

Get a profile’s channels

GET /api/voice/profiles//channels Return the channel ids a profile is assigned to.

string

required

The profile id.

Response

{ "channel_ids": ["chan_a", "chan_b"] }

Set a profile’s channels

PUT /api/voice/profiles//channels Replace the set of channels assigned to a profile.

string

required

The profile id.

string[]

required

The complete list of channel ids to assign.

Response

{ "message": "Profile channels updated successfully" }

Compose in character

If a profile has a personality, Voice Studio can generate a fresh in-character utterance you can edit before speaking — this is what the Compose button in the Generation tab produces.

Compose an utterance

POST /api/voice/profiles//compose Produce a fresh utterance in the profile’s character voice. Takes no body.

string

required

The profile id. The profile must have a personality set.

string

The generated in-character utterance.

string

The size of the model that produced the text.

curl -X POST https://app.peachdeskai.com/api/voice/profiles/prof_123/compose \
  -H "X-API-Key: $PEACHDESK_API_KEY"

Response

{ "text": "Hey there! Always happy to help — what can I do for you?", "model_size": "0.6B" }

There is no standalone rewrite endpoint. To rewrite text in character during synthesis, call the generation endpoint with personality=true instead.

Import & export

Profiles can be moved between workspaces as a single ZIP archive that bundles the profile metadata and its samples.

Export a profile

GET /api/voice/profiles//export Download a profile as a ZIP archive (application/zip). The response is a file attachment named profile-<name>.peachdesk.zip.

string

required

The profile id.

curl https://app.peachdeskai.com/api/voice/profiles/prof_123/export \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -o support-agent.zip

Import a profile

POST /api/voice/profiles/import Create a profile from a previously exported ZIP archive. Request is multipart/form-data.

file

required

The exported ZIP archive. Max 100 MB.

Returns the newly created Voice Profile object. Files over 100 MB return 400.

curl -X POST https://app.peachdeskai.com/api/voice/profiles/import \
  -H "X-API-Key: $PEACHDESK_API_KEY" \
  -F "file=@support-agent.zip"

Resources

Authentication & Errors

Voice Engine API

Bibha BRAHMA LLM

Bibha VAANI TTS

Bibha LIPI STT

Voice Profiles & Cloning

Object: Voice Profile

Managing profiles

Create a profile

List profiles

Get a profile

Update a profile

Delete a profile

Built-in voices

List built-in voices for an engine

List the curated built-in catalog

Preview a built-in voice

Reference samples (cloning)

Add a sample

List samples

Update a sample’s transcript

Delete a sample

Avatars

Upload an avatar

Get an avatar

Delete an avatar

Default effects

Set or clear the default effects chain

Channel assignment

Get a profile’s channels

Set a profile’s channels

Compose in character

Compose an utterance

Import & export

Export a profile

Import a profile

​Object: Voice Profile

​Managing profiles

​Create a profile

​List profiles

​Get a profile

​Update a profile

​Delete a profile

​Built-in voices

​List built-in voices for an engine

​List the curated built-in catalog

​Preview a built-in voice

​Reference samples (cloning)

​Add a sample

​List samples

​Update a sample’s transcript

​Delete a sample

​Avatars

​Upload an avatar

​Get an avatar

​Delete an avatar

​Default effects

​Set or clear the default effects chain

​Channel assignment

​Get a profile’s channels

​Set a profile’s channels

​Compose in character

​Compose an utterance

​Import & export

​Export a profile

​Import a profile

Object: Voice Profile

Managing profiles

Create a profile

List profiles

Get a profile

Update a profile

Delete a profile

Built-in voices

List built-in voices for an engine

List the curated built-in catalog

Preview a built-in voice

Reference samples (cloning)

Add a sample

List samples

Update a sample’s transcript

Delete a sample

Avatars

Upload an avatar

Get an avatar

Delete an avatar

Default effects

Set or clear the default effects chain

Channel assignment

Get a profile’s channels

Set a profile’s channels

Compose in character

Compose an utterance

Import & export

Export a profile

Import a profile