ElevenLabs Speech To Speech V2
Neural Network
ElevenLabs Speech To Speech V2 is a speech conversion model with maximum control over content and prosody.
Specifications
- Input
- audio
- Output
- audio
Prompt cost
(per 1M tokens)
How it works ElevenLabs Speech To Speech V2?
ElevenLabs Speech To Speech V2 is a speech conversion model from ElevenLabs for tasks requiring maximum control over content and output prosody. You provide a source audio track and receive new speech where intonation, pace, and accents are aligned as required by the script. Through BotHub, the model opens in the same window as other neural networks in the catalog: you can switch without new integration, access is provided without VPN or foreign cards, payment is in rubles based on usage, tokens do not expire, and connection to multimodal systems is via a unified OpenAI-compatible API. For teams, there are corporate terms with contracts, invoices, and an admin panel with limits; data is transmitted via AES-GCM encryption and is not saved. Scenarios include: re-voicing commercials with precise delivery, dubbing and video adaptation, intonation alignment in course narration, refining character lines in games, and quick iterations when comparing multiple versions of a phrase.Frequently asked questions about ElevenLabs Speech To Speech V2
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
Eleven English v2 is a speech conversion model: you provide an audio recording, and you get new audio in the desired voice as output. It is suitable for voiceovers, dubbing, timbre changes, and working with delivery when preserving the intonation and rhythm of the original recording is more important than synthesizing speech from scratch. Available via BotHub without VPN or foreign cards.
This is not about text, but audio: the model accepts and outputs sound. According to our data, the context is up to 5000 tokens, with the same amount allocated for a single response. This is enough for short lines and voiceover fragments; longer recordings are better split into parts and processed sequentially.
In our data, reasoning before answering is not noted for this model. Eleven English v2 is a sound processing tool: it converts provided audio into new audio, so thinking and text verification stages are not part of its scenario. For tasks requiring a chain of thought, you can switch to a text model in BotHub.
Function calling and structured JSON output are not noted for Eleven English v2 in our data. This is expected: the model works with audio and outputs sound, not software structures. If your scenario requires function calling or JSON responses, BotHub has text models with these capabilities.
From this list, the model accepts audio — that is what its entire operation is based on: input and output are sound. There are no notes about images or video in our data, so do not count on them. If you need to upload a picture or a clip, BotHub has a model for that task.
There is no information in our data about the languages the model works with, so we will not promise quality in Russian. The most honest way to check is to upload a short fragment of your recording and evaluate the result by ear. Payment in BotHub is based on usage, so such a test will be inexpensive.