NVIDIA released Magpie TTS Multilingual on August 10, 2026 — a 364M-parameter open-weights text-to-speech model covering 12 languages, distributed under Apache 2.0 with a matching NIM container for on-prem serving. The key metric: 32ms Time to First Audio on a single B200. That's the entire TTS budget before ASR and LLM respond, leaving room for a sub-200ms end-to-end pipeline.
The language roster spans English, Spanish, French, German, Italian, Vietnamese, Mandarin, Hindi, Japanese, Modern Standard Arabic, Korean, and Brazilian Portuguese. Each language ships male and female voices through a shared multilingual representation, eliminating 12 separate models. Code-switching for Hindi and Japanese uses IPA grapheme-to-phoneme processing and custom pronunciation dictionaries — essential when users speak product or person names mid-sentence in a second script.
Two architectural choices drive the latency. First, frame stacking: the decoder predicts two audio frames per step instead of one, cutting iterations in half. Second, a local transformer preserves dependency relationships between simultaneously generated codebook tokens. Stacking alone would collapse audio quality; together they deliver throughput without the penalty.
Benchmark figures from NVIDIA TTS NIM Performance v26.07, measured on-prem across three trials. Single concurrent stream: B200 hits 32ms TTFA at 12.1× real-time throughput; H100 at 47ms / 14.7×; DGX Spark at 53ms / 9.8×; A100 at 79ms / 12.2×. At 64 concurrent streams, B200 reaches 239ms TTFA but sustains 319.81× real-time throughput — generating audio 300+ times faster than playback. DGX Spark at 64 streams climbs to 962ms TTFA while throughput drops to 75.88×, reflecting the edge-class hardware ceiling and exposing where Spark belongs: dev and low-concurrency edge deployments, not high-fan-out serving.
| Hardware | Concurrent Streams | TTFA (ms) | Real-time Throughput |
|---|---|---|---|
| B200 | 1 | 32 | 12.1× |
| H100 | 1 | 47 | 14.7× |
| DGX Spark | 1 | 53 | 9.8× |
| A100 | 1 | 79 | 12.2× |
| B200 | 64 | 239 | 319.81× |
| DGX Spark | 64 | 962 | 75.88× |
The NIM container and Hugging Face checkpoint share identical weights. The NIM provides the tuned serving stack that delivers production latencies; the HF checkpoint enables fine-tuning and research. Apache 2.0 eliminates per-call licensing costs and data-egress fees — the core argument for self-hosted over managed APIs. Magpie removes both friction points: enforced data residency and predictable cost at scale.
NVIDIA positions this against integrated speech models (audio in / out, single API call) and cascaded architecture (separate ASR, TTS, LLM). The tradeoff matters: integrated models simplify wiring; cascaded stacks enable independent upgrade paths, per-component latency visibility, and LLM swaps without re-evaluating TTS. Magpie targets cascaded deployments — it assumes ASR and LLM are present and TTS must not consume the remaining latency budget.
| Dimension | Integrated (audio in/out) | Cascaded (ASR + LLM + TTS) |
|---|---|---|
| Wiring complexity | Single API call | Multiple components to wire |
| Upgrade path | Coupled — full model swap | Independent per component |
| Latency visibility | Opaque end-to-end | Per-component measurement |
| LLM swap cost | Re-evaluate entire model | Swap LLM without re-evaluating TTS |
| Magpie fit | Not targeted | Primary deployment target |
Gaps not addressed: fine-tuning recipes for custom voices, memory footprint at serving time, or latency numbers for quantized variants. For ARM edge devices or consumer GPUs, the DGX Spark 962ms / 64-stream ceiling is real, not an outlier. Plan accordingly.
If you're running multilingual voice agents on H100 or B200 infrastructure and currently pay per-call for managed TTS, Magpie's 32–47ms single-stream TTFA and Apache 2.0 license make the migration economics straightforward.