Subtitle Translator

⚠️ Adjusting these settings can affect performance, cost, and translation quality. Proceed with caution.

Gemini Model:

Select the AI model for translation.

Temperature:

Controls randomness (0.0-2.0). Higher values = more creative/random.

Top-P:

Nucleus sampling (0.0-1.0). Considers tokens comprising the top P probability mass.

Top-K:

Sample from the K most likely tokens (integer >= 1).

Max Output Tokens:

Maximum number of tokens to generate per request (check model limits).

Stop Sequences (comma-separated):

Stop generation if these strings appear. Leave blank if none.

Use Proxy

Enable if direct Gemini API access is blocked (e.g., sanctioned regions).

Base Delay (ms):

Delay between successful chunk requests.

Quota Delay (ms):

Delay after hitting a rate limit (429 error).

Number of Chunks:

Split SRT into this many parts (1-100).

System Prompt / Instructions:

Instructions for the AI model.