Skip to content
KugelAudioDocs

Voice Cloning

Clone a voice from reference audio with KugelAudio: audio requirements, speaker consent, creating clones in the dashboard or SDK, and managing cloned voices.

On this page

Voice cloning lets you create a synthetic voice from reference audio.

How It Works

  1. Upload reference audio - Provide clean, representative speech
  2. Processing - Our AI analyzes the voice characteristics
  3. Voice created - Use your new voice in any TTS request

Requirements

Audio Quality

For best results, your reference audio should be:

  • Format: WAV, MP3, OGG, M4A, or FLAC
  • File size: At most 50 MiB per reference
  • Channels: Mono preferred
  • Quality: Clean, no background noise

Content Guidelines

✅ Good audio:

  • Clear speech with natural pacing
  • Single speaker only
  • Minimal background noise
  • Natural emotional range
  • Free of filler words (um, uh, ah, hmm) unless you want them in the output

❌ Avoid:

  • Multiple speakers
  • Background music
  • Heavy reverb or echo
  • Whispered or shouted speech
  • Heavily compressed audio
  • Recordings with frequent filler sounds or hesitations
  • Long gaps or extended silence between sentences, unless you want the cloned voice to reproduce those pauses

A voice is cloned only with the speaker's consent. Every request that creates a voice or adds reference audio to one must state:

I confirm that I am the speaker of this recording or that I have the speaker's permission to create a synthetic voice from it.

In the dashboard this is a checkbox beside your recording. In the API and the SDKs it is the required field speaker_consent, which must be true:

  • A request without it is rejected with 400 and error_code SPEAKER_CONSENT_REQUIRED. No voice and no audio are stored.
  • We record the confirmation with your account, the API key if one was used, the voice and the time.
  • A voice without a recorded confirmation cannot be published (409, SPEAKER_CONSENT_MISSING). Voices created before this requirement have none: add their reference audio again with speaker_consent set to true. They keep working for synthesis.

Report misuse

If a voice made with KugelAudio impersonates someone, was cloned without the speaker's consent, or is used for fraud or deception, report it at kugelaudio.com/en/report-abuse. Anyone can report, with or without an account. You get a reference for the report, and we reply to the email address you give.

Creating a Voice Clone

Via Dashboard

  1. Go to Dashboard → Voices → Create Voice
  2. Upload your reference audio and confirm the speaker consent
  3. Enter a name and description
  4. Click Create Voice
  5. Wait for processing to finish

Via SDK

Set supported_languages to the languages the voice speaks. The first entry is the voice's primary language: when a synthesis request omits language, it normalizes numbers and dates in text whose language is not recognized (Text normalization). It defaults to ["en"], so set it for any non-English voice.

Python
from kugelaudio import KugelAudio

client = KugelAudio(api_key="YOUR_API_KEY")

# Create a voice with reference audio
voice = client.voices.create(
    name="My Custom Voice",
    sex="female",
    speaker_consent=True,  # required, see "Speaker consent"
    description="Cloned from reference audio",
    category="conversational",
    supported_languages=["de"],  # first entry = primary language
    reference_files=["reference.wav"],
)

print(f"Created voice: {voice.id}")
print(f"Name: {voice.name}")

For Java, see SDK voices.

Using Cloned Voices

Once created, use your cloned voice like any other. The examples continue from the create step (voice.id); in cURL, set VOICE_ID to the id from the create response.

Python
# Use your cloned voice
audio = client.tts.generate(
    text="Hallo, hier spricht meine geklonte Stimme!",
    model_id="kugel-3.5",
    voice_id=voice.id,
    language="de",
)

audio.save("cloned_output.wav")

Best Practices

Optimizing Voice Quality

Use high-quality source audio

The quality of your cloned voice depends heavily on the source audio. Use professional recordings when possible.

Provide diverse samples

Include a range of intonations, emotions, and sentence types in your reference audio for a more natural clone.

Adjust CFG scale

Experiment within the supported cfg_scale range of 1.2 to 2.5.

Remove filler sounds from samples

If your output contains unwanted "um"s, "ah"s, or hesitations, re-record or edit your reference audio to remove them. The model faithfully reproduces what it hears in the samples: clean input produces clean, controllable output. You can always add fillers via your text prompts later.

Troubleshooting

IssueSolution
Voice sounds roboticUse higher quality source audio, try lower CFG scale
Voice sounds differentEnsure source audio is clean, try different text samples
Accent not preservedInclude more diverse samples, use longer reference audio
Inconsistent outputTry a different CFG value within 1.2–2.5
Unwanted filler sounds (um, ah, hmm)Re-record or edit reference audio to remove fillers. See Content Guidelines
Unexpected long pausesRe-record or edit reference audio to remove long gaps between sentences: the model can learn and reproduce these pauses

Managing Cloned Voices

The examples below act on your own voice: voice.id from the create step, or $VOICE_ID in cURL.

List Your Voices

Python
result = client.voices.list()

for voice in result.voices:
    print(f"{voice.id}: {voice.name}")

Update Voice

Python
updated = client.voices.update(
    voice_id=voice.id,
    name="Updated Name",
    description="Updated description",
)
print(f"Updated: {updated.name}")

Delete Voice

Python
client.voices.delete(voice_id=voice.id)

Managing Reference Audio

You can add and remove reference audio files after creating a voice.

List References

Run this right after creating a voice to confirm which reference files were accepted.

Python
for ref in client.voices.list_references(voice.id):
    print(f"{ref.id}: {ref.name}")

Add Reference

Python
ref = client.voices.add_reference(
    voice_id=voice.id,
    file="new_reference.wav",
    speaker_consent=True,
    reference_text="Optional transcript of the audio.",
)
print(f"Added reference: {ref.id}")

Delete Reference

Python
client.voices.delete_reference(voice_id=voice.id, reference_id=ref.id)

Publishing Voices

Request that your voice be made public. It will be marked as pending verification until reviewed by an admin.

Python
published = client.voices.publish(voice_id=voice.id)
print(f"Pending verification: {published.pending_verification}")

Generating Voice Samples

Trigger sample audio generation after uploading at least one reference:

Bash
curl -X POST "https://api.kugelaudio.com/v1/voices/$VOICE_ID/generate-sample" \
  -H "Authorization: Bearer $KUGELAUDIO_API_KEY"

The response contains sample_s3_path and a signed sample_url.

AI Transparency & Watermarking

Voice-cloned output uses the same in-band watermark and HTTP disclosure header as every other synthesis response. See AI-generated audio marking for the wire details, detector example, and limitations.

Privacy & Ethics

Guidelines

  1. Get consent - Always obtain permission before cloning someone's voice; every clone request records your speaker consent
  2. Disclose synthetic speech - Be transparent when using cloned voices in public-facing contexts
  3. No impersonation - Don't use cloned voices to deceive or defraud
  4. Respect rights - Don't clone voices of public figures without authorization

Next Steps