gianpaj/cog-orpheus-3b-0.1-ft 📝🔢❓ → 🖼️

▶️ 723 runs 📅 Apr 2025 ⚙️ Cog 0.14.3 🔗 GitHub
text-to-speech voice-cloning

Performance

7.4sTypical run time
~79sCold start (first call)
723Total runs

About

Spanish and English Text to Speech model from Canopy Labs (3b-es_it-ft-research_release)

Example Output

Output

Example output

Performance Metrics

7.36s Prediction Time
79.46s Total Time
All Input Parameters 
{
  "text": "Hola, me llamo Javi, encantado de conocerte <giggle>",
  "top_p": 0.95,
  "voice": "javi",
  "temperature": 0.6,
  "max_new_tokens": 1200,
  "repetition_penalty": 1.1
}
Input Parameters
text (required) Type: string
Text to convert to speech
top_p Type: numberDefault: 0.95Range: 0.1 - 1
Top P for nucleus sampling
voice Default: javi
Voice to use
temperature Type: numberDefault: 0.6Range: 0.1 - 1.5
Temperature for generation
max_new_tokens Type: integerDefault: 1200Range: 100 - 2000
Maximum number of tokens to generate
repetition_penalty Type: numberDefault: 1.1Range: 1 - 2
Repetition penalty
Output Schema

Output

Type: string • Format: uri

Example Execution Logs 
Setting `pad_token_id` to `eos_token_id`:128258 for open-end generation.
Version Details
Version ID
666dc0c400952f2c18f0a46233dca2053ebef622754769878cd5497e20714650
Version Created
April 11, 2025
Run on Replicate →