Audio & Speech

Today's Stories

1. 1

   Paris AI voice startup Gradium raises $100M seed, expands to Bay Area

   Gradium, a Paris-based startup building voice AI models, closed its seed round at $100 million（約160億円） total, adding Nvidia to its investor roster after initially raising $70 million（約110億円） in December. The company is opening a Bay Area office to strengthen its talent position. Gradium competes in a crowded voice AI market alongside ElevenLabs (valued at $11 billion（約1.8兆円） in February) and major players like Google's Gemini. The startup has already landed major customers including French auto manufacturer Renault since launching in December, suggesting its technology for delivering voice at scale with ultra-low latency is gaining real commercial traction.

   The company is using fresh funding to establish itself near major AI labs in the Bay Area—a strategic bet that proximity to Anthropic, Google, Meta, and OpenAI is essential for competing in AI talent markets, even though Gradium's founders and initial backers were rooted in Paris's own AI hub.

2. 2

   Cohere releases open-source Arabic speech-to-text model

   Cohere released Cohere Transcribe Arabic, a 2-billion-parameter open-source model for Arabic speech recognition. The model is available on Hugging Face and through the Cohere API under the Apache 2.0 license. According to Cohere, it is the most accurate open-source Arabic speech-to-text system available and outperforms Whisper Large V3 and the standard Cohere Transcribe model in benchmarks. It addresses Arabic's specific challenges—dialect variety, bilingual Arabic-English conversations, code-switching, and specialized vocabulary—which are difficult for general speech recognition systems to handle accurately.

   Human ratings on a 1–5 scale show Cohere Transcribe Arabic outperforms both Whisper Large V3 and the standard Cohere Transcribe model in overall quality, dialect faithfulness, and code-switching. The model is available now on Hugging Face and via the Cohere API.

3. 3

   Anthropic clashes with Trump's White House, rejects Washington playbook

   The Trump administration has twice taken actions against Anthropic—labeling it a "supply chain risk" in April after the company refused Pentagon contract language, and imposing export controls on its Mythos and Fable AI models two weeks ago following discovery of a jailbreak. OpenAI, by contrast, announced it was withholding release of GPT-5.6 at the U.S. government's request on the same day those controls were relaxed. Anthropic, valued at $965 billion（約150兆円） and preparing for an IPO expected in the coming months, has refused the flattery, donations, and appointment of Trump allies that other tech giants (Meta, Amazon, Apple) and OpenAI have used to stay in the administration's favor. Trump administration officials have publicly attacked CEO Dario Amodei as a "liar" with a "God-complex" and an "ideological lunatic," and accused the company of "regulatory capture." Continued hostility could make it harder to sell public market investors on the stock listing and significantly hamper the company's ability to develop advanced AI models.

   Anthropic CEO Amodei reportedly called Trump "a feudal warlord" in a now-deleted Facebook post, and his sister and cofounder Daniela Amodei donated to Kamala Harris's campaign. Unlike OpenAI's policy chief Chris Lehane and cofounder Greg Brockman (the largest donor to Trump Super PAC MAGA Inc.), Anthropic has made no similar hires of Trump-aligned figures to its leadership.

4. 4

   Netflix uses AI-generated Gene Wilder voice for Wonka reality show

   Netflix is premiering Wonka's The Golden Ticket on September 23rd, a reality competition based on the fictional Wonka universe. The show's voiceover uses an AI-generated version of Gene Wilder's voice, created in partnership with AI audio company ElevenLabs and with consent from Wilder's family. This extends Netflix's pattern of using AI-generated celebrity voices for content—the company has previously recreated voices of Michael Caine and Stan Lee. For viewers, it means encountering synthetic versions of iconic figures in new productions, blurring the line between archival and synthetic media in mainstream entertainment.

   The two-part finale airs on September 30th. The show will feature 12 golden ticket winners and their chosen partners competing in a high-stakes social experiment, with one champion crowned by the end.

5. 5

   Reddit user struggles to replicate Pocket TTS model from paper without training code

   A developer attempting to implement Pocket TTS (a text-to-speech model from a published paper) is encountering significant difficulties because the authors have not released the training or fine-tuning code. The developer trained a smaller version on LJSpeech and LibriSpeech datasets, but the model fails to generate meaningful speech even for text it saw during training, despite low loss values. Attempts to fix the issue using scheduled sampling and noise injection have not worked. Reproducibility barriers in AI research can slow adoption and learning. When researchers publish papers without releasing training code, it forces others to reverse-engineer the implementation from scratch, introducing delays and increasing the risk of implementation errors that make the paper's results difficult to verify or build upon.

   The developer's core challenge appears to be a gap between low training loss and poor inference quality—a sign of possible exposure bias, mode collapse, or data/model mismatch. The community response to this post may surface whether other practitioners have faced similar issues or found workarounds for Pocket TTS implementation.

6. 6

   Ask HN: Best AI Voice Narration Tool Beyond ElevenLabs

   A Hacker News user asked the community whether there are AI voice narration tools better than ElevenLabs for character voice acting in animation, citing concerns that ElevenLabs produces flat, inconsistent character voices. The user's experience—spending $75 on a human voice actor and finding the work unusable, combined with limitations in existing free tools—highlights that current AI voice solutions may not yet reliably handle emotional tone and character consistency at scale, a gap that matters for creators on a budget.

   The question remains open on Hacker News with no answers logged; the community response will indicate whether a consensus SOTA (state-of-the-art) tool exists for this use case or whether the problem remains unsolved.