Audio & Speech

Today's Stories

1. 1

   Beatport bans AI-generated music from DJ marketplace

   Beatport now bans tracks made entirely or mostly by AI. The rule is enforced with a detection tool from partner Beatdapp that filters AI tracks during upload, and rights holders receive a notice when a track is rejected. A Beatport survey found 60 percent of users would not play AI music in their sets, and 77 percent prefer human-made music. CEO Matt Gralen stressed the difference between a tool that helps people and a system that replaces them entirely.

   Other platforms like Deezer have recently taken similar steps, with Deezer noting that nearly 50 percent of its daily uploads are AI-generated.

2. 2

   Japan to test AI for 119 emergency call handling

   The Fire and Disaster Management Agency plans to launch a model project in fiscal 2027 to use AI in handling 119 emergency calls. The AI-powered speech recognition system will automatically transcribe calls, and related expenses will be included in the fiscal 2027 budget request. The aim is to streamline the process of recording call details and ease the burden on staff amid a growing number of ambulance dispatches. The system will also advise staff on necessary actions based on each call's content.

   AI will also be introduced for the "#7119" telephone consultation service, and the agency plans to test wearable robots for rescue operations and drone-based AI fire detection. In July, the government released an interim report promoting vertical AI implementation in firefighting by 2030.

3. 3

   AWS Quick and fal unite for agentic creative workflows

   AWS announced an integration where Amazon Quick, an agentic AI workspace, connects to fal's generative media platform via Model Context Protocol (MCP). This enables creators to run workflows like an eight-panel storyboard or a music-video concept within one interactive session. Creative teams face demand that exceeds capacity—78% of creative leaders report this—and faster generation alone doesn't solve workflow fragmentation. The integration retains context across steps, adds human approval gates, and packages repeatable processes as shareable Skills.

   The setup requires a fal account and API key, plus permission to add a remote MCP connector in Amazon Quick. The storyboard workflow uses FLUX.1 Kontext for reference-guided panel generation.

4. 4

   Leafnet and BBIX team up on voice-AI services

   Leafnet and BBIX began collaborating in August 2026 to build a new service that combines voice and AI. Leafnet brings its voice-calling expertise, while BBIX adds its network and cloud platform technology. The partnership aims to make AI more accessible through natural voice interfaces, a field expected to grow as AI evolves. The two companies hope to expand voice communication possibilities in daily life and business.

   The new service will pair Leafnet's voice service with BBIX's network platform and cloud services. Details on pricing or availability have not been announced.

5. 5

   NVIDIA MPS slashes ASR GPU needs by 75% on AWS

   AWS, NVIDIA, and Heidi Health published a solution that cuts GPU infrastructure for automatic speech recognition (ASR) inference by 75 percent—from 16 GPU instances down to 4—using NVIDIA CUDA Multi-Process Service (MPS) with Triton Inference Server on Amazon EC2. A single ASR request typically uses only 15–20 percent of a GPU's compute capacity, leaving 80 percent idle due to default time-slicing. By partitioning the GPU into concurrent instances, the setup maintains sub-second latency at 92.1 requests per second (RPS) per GPU, which could reduce operational costs for high-throughput speech services.

   The solution is implemented on Amazon EC2 g6e.4xlarge and g7e.4xlarge instances (NVIDIA L40S, 48 GB) and requires NVIDIA drivers 535+ with CUDA 12.x. Code is available in an open-source repository, deployable with three commands.

6. 6

   Deepgram expands SageMaker AI observability metrics

   Deepgram has introduced two capabilities for its speech-to-text and text-to-speech models on Amazon SageMaker AI: Deepgram Enhanced Metrics for billing and usage transparency, and Prometheus and OpenTelemetry support for engine-level and per-GPU visibility. Both are available today on Deepgram SageMaker AI deployments. These innovations address the observability trade-off of self-hosted speech AI, giving you visibility into billing, feature usage, and engine behavior that was previously locked inside the vendor's container. They work within AWS Marketplace network isolation, so you can reconcile your AWS bill and monitor capacity without opening a network path or adding extra permissions.

   Detailed observability is on by default for newly created endpoints, publishing every 60 seconds. You can query the metrics with PromQL from CloudWatch, Grafana, or any Prometheus-compatible tool, and filter by endpoint, instance, or GPU.