The Best Deepgram Alternative for Real-Time Voice Agents

Compare Pulse STT by Smallest AI vs Nova 3 on latency, WER, language support, on-premise deployment, and pricing for real-time voice agents.

The Best Deepgram Alternative for Real-Time Voice Agents

Compare Pulse STT by Smallest AI vs Nova 3 on latency, WER, language support, on-premise deployment, and pricing for real-time voice agents.

Why teams switch to Pulse Speech to Text

Here's what that difference looks like in production.

Why teams switch to Pulse Speech to Text

Here's what that difference looks like in production.

Why teams switch to Pulse Speech to Text

Here's what that difference looks like in production.

Industry-lowest Word Error Rate

Lowest WER across 30+ languages-on real-world audio with noise, accents, and overlapping speakers.

Industry-lowest Word Error Rate

Lowest WER across 30+ languages-on real-world audio with noise, accents, and overlapping speakers.

Auto Language Detection & Code Switching

Identifies language and switches mid-sentence.

Auto Language Detection & Code Switching

Identifies language and switches mid-sentence.

Profanity Filtering & Word Boosting

Censor what shouldn't be there. Prioritise what should.

Profanity Filtering & Word Boosting

Censor what shouldn't be there. Prioritise what should.

Speech intelligence built in

Speaker diarization, real-time sentiment analysis, emotion detection, and automatic language identification

Speech intelligence built in

Speaker diarization, real-time sentiment analysis, emotion detection, and automatic language identification

Pulse vs Nova

A factual, model-level comparison on the metrics that matter in production.

Pulse vs Nova

A factual, model-level comparison on the metrics that matter in production.

Pulse vs Nova

A factual, model-level comparison on the metrics that matter in production.

Features

Features
Pulse
Nova 3
Time to First Transcript
70 ms
100ms
Streaming WER
5.42% avg WER (Open ASR, #2)
8.60% avg WER
Compliance
SOC 2, HIPAA, GDPR
SOC 2
Pricing
~$0.008/min
$0.0059/min

Benchmarks

Dataset
Benchmark
Pulse
Nova 3
Meeting recordings
AMI
7.32
17.04
Earnings calls
Earnings22
9.04
15.79
Podcasts & audiobooks
GigaSpeech
9.52
10.05
Clean audiobook speech
LibriSpeech clean
1.73
3.20
Noisy audiobook speech
LibriSpeech other
3.74
6.60
Financial speech
SPGISpeech
2.04
2.99
TED talks
TED-LIUM
3.68
3.59
Parliament & accented speech
VoxPopuli
6.32
9.55
Overall average
Average (8 datasets)
5.42
8.60
Public leaderboard rank
Open ASR rank
#2 (tied)
US English
FLEURS en_us
3.92%

Certified & Compliant

Guarding your data with enterprise security

Certified & Compliant

Guarding your data with enterprise security

Proactive Defense

Anticipating threats before they emerge, thanks to our advanced monitoring.

Proactive Defense

Anticipating threats before they emerge, thanks to our advanced monitoring.

Proactive Defense

Anticipating threats before they emerge, thanks to our advanced monitoring.

Frequently
asked questions

Can Pulse STT run on-premise?

Does Deepgram support emotion recognition?

How does Pulse STT's latency compare to Deepgram Nova-3?

What does switching from Deepgram to Pulse STT involve?

Build the future of voice agent orchestration

Build the future of voice agent orchestration

Build the future of voice agent orchestration