Limit 1
Google WaveNet has no Telugu voice, so Bharat Essential cannot serve te-IN and the benchmark uses Bharat Standard.
తెలుగు · measured India voice evidence
Telugu AI voice: runtime path, real pack lines and measured latency
Telugu resolves to Bharat Standard, Azure Speech voice en-IN-NeerjaIndicNeural, the te-IN dedicated pack and the cascade engine. Three fresh production-adapter runs measured a 608 ms median to the first non-empty 8 kHz PCM audio chunk on 15 Aug 2026.
Measured 15 Aug 2026 · three runs · 8 kHz PCM16 · no provider fallback
Which pack, engine and vendor voice serve Telugu?
The runtime trace starts with BaseBusiness.get_voice_pack, continues through voice_config.resolve_pack and ends at voice_config.resolve_engine. The generic integration returned none from the generic business integration, resolve_pack selected the te-IN dedicated pack, and resolve_engine selected cascade. The TTS factory then selected Azure Speech voice en-IN-NeerjaIndicNeural, displayed as Neerja Indic, while speech recognition remained Sarvam saaras:v3.
| Runtime check | Resolved value | Meaning |
|---|---|---|
| get_voice_pack | None from the generic business integration | A client-specific pack would override the generic registry result. |
| resolve_pack | te-IN dedicated pack | Fixed lines and language rules come from a locale-specific pack. |
| voice_config | Bharat Standard · Azure Speech · en-IN-NeerjaIndicNeural | The grade write-through pins the production TTS provider and voice identifier. |
| resolve_engine | cascade | The cascade engine uses separate STT, reasoning and TTS stages. |
| speech recognition | Sarvam saaras:v3 | The primary recognition locale is te-IN. |
The runtime table names the actual Telugu call path rather than a language-support logo: generic integration result, resolved pack, sold grade, provider voice, engine and speech-recognition provider. A client integration can replace the generic pack, but the benchmark fixture used no client override.
What first-audio latency did Telugu measure?
Telugu measured 577, 608, 823 ms across three isolated production-adapter requests. The median was 608 ms. Each run synthesized “నమస్కారం, మీకు ఎలా సహాయం చేయగలను?” at 8 kHz PCM16, started the timer immediately before speak(text), stopped at the first non-empty audio chunk, verified the configured adapter had not failed over, and cancelled the iterator after that first chunk.
| Run | First audio | Audio format | Adapter result |
|---|---|---|---|
| Run 1 | 577 ms | 8 kHz PCM16 | Azure Speech; no fallback |
| Run 2 | 608 ms | 8 kHz PCM16 | Azure Speech; no fallback |
| Run 3 | 823 ms | 8 kHz PCM16 | Azure Speech; no fallback |
The latency table reports a median of 608 ms from three measurements on 15 Aug 2026. The number is a reproducible TTS-adapter read, not end-to-end call latency, a percentile, or an uptime promise. Caller speech, transcription, model generation, carrier delivery and handset playback remain outside the timer.
What register does the resolved Telugu pack use?
The dedicated Telugu pack uses polite spoken Telugu with అండి as a respect marker. The recovery language is conversational rather than literary, and the agent persona uses feminine verb forms in its longer retraction line.
| Pack event | Exact runtime string | Register read |
|---|---|---|
| Filler while a tool runs | ఒక్క క్షణం అండి | A short polite spoken-Telugu hold line. |
| Retry after unclear speech | క్షమించండి అండి, ఒక్కసారి మళ్ళీ చెప్తారా? | Polite conversational retry with అండి. |
| Retract an unsupported action | క్షమించండి, అది నేను ఇక్కడ నుంచి పంపలేను. మీ నంబర్ తీసుకుంటాను, ఎవరైనా మీకు పంపిస్తారు. | A direct retraction followed by a human fallback. |
The sample table copies three fixed strings from the te-IN dedicated pack: the tool filler, unclear-speech retry and unsupported-action retraction. Telugu therefore has explicit runtime register evidence.
What are the honest Telugu limits?
Limit 2
Azure's native te-IN-ShrutiNeural and te-IN-MohanNeural voices were retired after the 2 Aug 2026 owner audition; runtime resolution rewrites either old selection to an Indic multilingual voice.
Limit 3
The production Telugu voice is en-IN-NeerjaIndicNeural reading Telugu script, not a native te-IN voice identifier.
Limit 4
The 608 ms median excludes the built-in Azure connection and model warm-up, which the production adapter performs during call setup before the measured turn.
Limit 5
The first-audio benchmark measures TTS adapter latency only; it excludes caller speech recognition, model reasoning, carrier transit and handset playback.
Which evidence supports the Telugu hub?
The fresh GSC pull found 27 combined impressions across the Telugu cost page and the Telugu–Hindi demand page, so Telugu follows Hindi in the language queue. The existing language article remains the workflow and cost depth page; the multilingual category index provides cross-language buying context; and the product page describes the broader scheduled-language surface.
Other measured Indian-language hubs
The related hubs keep one route per language and never multiply languages by verticals. Every page publishes its own runtime trace, three dated latency reads and fixed-pack evidence.
Telugu voice questions
The answers preserve the runtime resolution, measured median, exact run values and pack limitation as standalone passages that do not depend on the tables.
Telugu resolves to Bharat Standard, the Azure Speech en-IN-NeerjaIndicNeural voice, Sarvam saaras:v3 recognition and the cascade engine. The generic integration returned no client-specific pack, so resolve_pack selected the te-IN dedicated pack on 15 Aug 2026.
Telugu measured 608 ms median first-audio latency across three fresh 8 kHz PCM16 adapter runs: 577, 608, 823 ms. The timer started immediately before the production adapter's speak call and stopped at the first non-empty audio chunk on 15 Aug 2026.
Telugu uses the te-IN dedicated pack. The fixed pack strings shown on this page are copied from the runtime object rather than written as marketing examples.