ElevenLabs API

Converts input text to natural-sounding speech audio using one of three specialized models (Eleven v3 for emotional expressiveness, Multilingual v2 for stability on long-form content, or Flash v2.5 for low-latency production). The system processes text character-by-character with per-character credit consumption (1 credit per character for standard models, 0.5-1 for Flash variants), respecting model-specific input limits (5k-40k characters) and language coverage (29-70+ languages). Output is streamed or returned as PCM audio at 44.1kHz with quality tiers from 128kbps (free/starter) to 192kbps (pro+).

Enables users to clone a voice from audio samples (instant cloning) or create a professional voice clone with higher fidelity through a managed process. Instant Voice Cloning (Starter tier+) accepts short audio samples and generates a cloned voice usable immediately in TTS synthesis. Professional Voice Cloning (Creator tier+) involves a more rigorous process with quality assurance, producing voices suitable for commercial use. Both methods integrate with the standard TTS pipeline, allowing cloned voices to be used across all three TTS models with the same character-based credit consumption.

Provides qualifying startups with 12 months of free access plus 33 million characters of free TTS credits (equivalent to ~33,000 minutes of audio). The program is designed to enable early-stage companies to build voice features without upfront costs. Eligibility criteria and application process are not fully documented. Grants are distributed through the ElevenLabs website or partner programs (Y Combinator, Techstars, etc.).

Enables team collaboration through workspace management with role-based access control and seat allocation. Different pricing tiers provide different numbers of workspace seats: Scale tier includes 3 seats, Business tier includes 10 seats, and Enterprise tier includes custom seat allocation. Seats enable multiple team members to access the same workspace, projects, and voice library. The system supports consolidated billing and team-level usage tracking. Workspace features include project organization, shared voice library access, and collaborative content creation.

Modifies voice characteristics (pitch, speed, tone, accent) of existing audio recordings through neural voice transformation, enabling voice customization without re-recording or voice cloning. The voice changer applies learned transformations to match target voice characteristics while preserving original speech content and intelligibility, suitable for accessibility adjustments, creative effects, and voice personalization.

Implements a credit-based pricing model where each API operation consumes credits based on input size and operation type (1 character = 1 credit for standard TTS, 0.5-1 credit per character for Flash models depending on tier). Credits are allocated monthly per subscription tier (10k-6M credits/month), with unused credits rolling over for up to 2 months, enabling cost predictability and budget management. Developers can monitor credit consumption per request and optimize usage patterns to reduce costs.

Maintains a persistent voice library where cloned voices, designed voices, and pre-built voices are stored as reusable profiles with unique identifiers. Developers can create, organize, and manage voice profiles across projects, enabling consistent voice usage across multiple synthesis requests without re-cloning or re-designing. Voice profiles support metadata tagging and organization, facilitating voice discovery and reuse at scale.

Generates speech and text content across 29-90+ languages depending on operation (TTS supports 29-70+ languages, STT supports 90+ languages), with automatic language detection for input content. The system automatically selects appropriate language-specific models and processing pipelines based on detected language, enabling seamless multilingual workflows without explicit language specification. Supports language mixing in some contexts (e.g., code-switching in dialogue).

Generates synthetic voices from natural language descriptions without requiring audio samples. Users provide text descriptions of desired voice characteristics (e.g., 'warm, deep male voice with slight accent'), and the system generates a unique voice that matches the description. The generated voice is assigned a voice ID and can be used immediately in TTS synthesis across all three TTS models, consuming standard per-character credits. This capability abstracts away the need for voice cloning from samples and enables rapid voice creation for diverse character types.

Transcribes audio in 90+ languages to text using Scribe v2 (batch/offline) or Scribe v2 Realtime (real-time streaming). The system performs automatic language detection, word-level timestamp generation, speaker diarization (identifying and separating up to 32 speakers), entity detection (up to 56 entity types), and dynamic audio tagging. Batch processing is optimized for long-form content; realtime processing achieves ~150ms latency (excluding network). Keyterm prompting (up to 1,000 custom terms) enables domain-specific vocabulary recognition. Output includes structured JSON with timestamps, speaker labels, and confidence scores.

Provides two dubbing modes: Automatic Dubbing (available Starter tier+) automatically translates and re-voices video content in target languages using TTS synthesis, and Dubbing Studio (available Starter tier+) offers a web-based editor for manual control over translation timing, voice selection, and lip-sync adjustments. Enterprise tier includes fully managed dubbing with Productions, where ElevenLabs handles the entire workflow. The system preserves original video timing, generates translated speech in target language voices, and optionally applies lip-sync adjustments. Dubbing integrates with the voice library and voice cloning capabilities, enabling brand-consistent dubbing across multiple languages.

Implements a credit-based billing system where users purchase monthly credit allowances (10k free, 30k-6M+ paid tiers) and consume credits per operation: 1 credit per character for standard TTS models, 0.5-1 credit per character for Flash models, and per-second rates for other operations (STT, dubbing, music/sound generation). Unused credits roll over up to 2 months with active paid subscription. Extra credits can be purchased at tier-specific rates ($0.36/minute free tier, $0.17/minute pro tier). The model enables predictable monthly costs while allowing flexibility for variable usage patterns.

Provides access to a curated library of 10,000+ pre-built synthetic voices across diverse characteristics (age, gender, accent, tone, emotion). Users can browse and select voices from the library for immediate use in TTS synthesis without cloning or design. Voice Remixing capability (details not fully documented) enables blending or modifying existing voices to create variations. All library voices integrate seamlessly with TTS models (v3, Multilingual v2, Flash v2.5) and consume standard per-character credits. The library is continuously expanded and updated.

Enables streaming of synthesized audio in real-time, allowing playback to begin before the entire audio is generated. The system streams audio chunks over HTTP or WebSocket (implementation details not fully documented) with Flash v2.5 model achieving ~75ms latency (excluding network/app overhead). Streaming is compatible with all TTS models and voice options. The capability supports progressive audio playback, enabling interactive applications (voice assistants, real-time dialogue systems) and reducing perceived latency for end-users.

Supports SSML (Speech Synthesis Markup Language) or similar markup for fine-grained control over pronunciation, emphasis, pacing, and prosody in synthesized speech. Users can annotate text with markup tags to control how specific words or phrases are pronounced, emphasize certain words, adjust speaking rate, and control intonation. The system parses markup and applies the specified prosody modifications during synthesis. This capability enables precise control over speech output for specialized use cases (medical terminology, proper nouns, emotional emphasis).

Enables synthesis of multi-speaker dialogue where different speakers are assigned different voices and the system maintains speaker consistency and timing alignment. Forced Alignment capability (details not fully documented) ensures that synthesized speech aligns with original timing or specified timing constraints, useful for dubbing or dialogue synchronization. The system processes dialogue with speaker labels, assigns voices per speaker, and generates synchronized audio output. This capability supports interactive narratives, audiobooks with multiple characters, and dubbed content.