Changelog

Latest updates, new features, and improvements to the Nur API.

v2.4.0January 15, 2026

Music Generation GA

  • NewGeneral availability of the /v1/music/generate endpoint for AI-powered music creation.
  • NewGenre control parameter supporting 9 genres: pop, rock, jazz, classical, electronic, hip-hop, ambient, lo-fi, and cinematic.
  • NewStems separation API to extract individual instrument tracks from generated music.
  • ImprovedMusic generation quality improved with higher fidelity output at 48kHz sample rate.
  • FixedResolved an issue where music tracks shorter than 15 seconds would occasionally contain silence at the end.
v2.3.0December 8, 2025

Voice Agents Tools

  • NewFunction calling support for voice agents, allowing agents to invoke external tools and APIs during conversations.
  • NewSession history API to retrieve full transcripts and tool call logs from past agent sessions.
  • NewMulti-language agent support enabling a single agent to detect and respond in the caller's language automatically.
  • ImprovedAgent response latency reduced by 40% with optimized inference pipeline.
  • FixedFixed a bug where agent sessions would not properly clean up after idle timeout.
v2.2.0November 12, 2025

AI Dubbing

  • NewLip-sync support for video dubbing that matches mouth movements to the translated audio.
  • NewAdded 40 new target languages for dubbing, bringing the total to 75 supported languages.
  • NewQuality scoring endpoint that returns a confidence score for each dubbed segment.
  • NewBatch processing support to submit multiple dubbing jobs in a single API call.
  • ImprovedSpeaker diarization accuracy improved by 25% for multi-speaker content.
v2.1.0October 3, 2025

Audio Enhancement

  • NewNew /v1/audio/enhance endpoint for AI-powered audio post-processing.
  • NewSilence removal filter to automatically trim dead air from recordings.
  • NewLoudness targeting to normalize audio to broadcast standards (LUFS targeting).
  • ImprovedNoise reduction algorithm upgraded to handle complex background environments like cafes and offices.
  • FixedFixed audio clipping artifacts that could occur with high-gain enhancement settings.
v2.0.0September 1, 2025

V2 API Launch

  • NewCompletely redesigned SDK with a more intuitive, resource-oriented interface.
  • NewStreaming support for TTS and STT endpoints with real-time chunk delivery.
  • NewWebhook system for receiving real-time notifications on long-running operations.
  • NewGranular auth scopes for API keys, enabling least-privilege access control.
  • BreakingResponse format changed: all responses now use a consistent envelope structure with top-level data key.
  • BreakingDeprecated v1 voice IDs. Existing v1 IDs will continue to work until March 2026 but should be migrated.
  • ImprovedAPI response times improved by 30% across all endpoints.
v1.5.0August 18, 2025

Voice Changer

  • NewVoice changer endpoint with 7 built-in effect presets: deep, bright, whisper, robot, echo, radio, and megaphone.
  • NewReal-time voice changing mode via WebSocket connection for live audio transformation.
  • NewIntensity control parameter (0.0 to 1.0) to fine-tune the strength of voice effects.
  • ImprovedVoice effect processing latency reduced to under 50ms for real-time applications.
v1.4.0July 22, 2025

Conversational AI

  • NewReal-time conversational AI endpoint — speak naturally and receive instant AI voice responses.
  • NewContext-aware dialogue with full conversation history and memory across turns.
  • NewStreaming mode for conversations with sub-300ms turn latency for interactive use cases.
  • ImprovedVoice activity detection (VAD) accuracy improved by 15% for cleaner turn-taking.
  • FixedFixed an issue where long conversation sessions would cause timeout errors.
v1.3.0June 10, 2025

Voice Cloning Improvements

  • ImprovedMinimum audio requirement reduced from 60 seconds to 30 seconds for voice cloning.
  • NewMulti-language cloning: a single voice clone can now generate speech in any of the 35 supported languages.
  • ImprovedHigher fidelity clones with improved prosody matching and natural breathing patterns.
  • FixedResolved cloning failures that occurred with certain MP3 encoding bitrates.
  • FixedFixed inconsistent voice quality when cloning from stereo vs. mono audio sources.

V1 to V2 Migration

If you are migrating from the V1 API, the primary changes involve the new response envelope format and updated voice IDs. Below is a summary of the key differences.

1// V1 response format (deprecated)
2{
3 "audio_url": "https://cdn.nur.ai/audio/abc123.mp3",
4 "duration": 4.2,
5 "voice_id": "old_rachel"
6}

V1 Deprecation Timeline

The V1 API will remain functional until March 2026. V1 voice IDs will be automatically mapped to their V2 equivalents during this transition period. After March 2026, V1 endpoints will return 410 Gone responses. We recommend migrating to V2 as soon as possible.