Implementing Synthflow for No-Code Voice AI Agents

You spend weeks building a custom voice agent—writing code, tuning NLP, connecting telephony. Then you discover a 30% booking conversion rate and latency spiking above 2 seconds at peak. Synthflow solves these problems out of the box for building voice AI agents. It is a no‑code/low‑code platform with a visual scenario builder, built‑in CRM integration, and ready templates for typical dialogues. We have implemented Synthflow for 15+ projects—from retail to healthcare—and the average time to launch the first agent was two days. Below we break down how the implementation works and what results it delivers. On average, development budget savings reach $15,000 by eliminating the need to hire a developer team. For a typical project, custom development would cost $30,000–$50,000, while Synthflow implementation costs $15,000 or less.

Synthflow vs Alternatives

From our data, Synthflow deploys twice as fast as VAPI and is up to three times cheaper than custom development. Clients save up to 60% of the development budget, which in monetary terms can be up to $15,000 per project.

How Synthflow Speeds Up Voice Agent Launch

Thanks to the visual builder and ready templates, the first agent launches in 1–3 days. CRM and telephony integration is done via API or built‑in connectors, eliminating manual setup. The platform automatically handles speech recognition, TTS, and call‑peering management. We guarantee stable operation with an SLA of 99% uptime.

Why Synthflow Is More Cost‑Effective Than Custom Solutions

Building a comparable voice agent on VAPI or a custom stack takes from two weeks with a budget 3–5 times higher. Synthflow provides a ready infrastructure: automatic speech recognition, TTS, call‑peering management. Our certified specialists have deployed the platform for 15+ projects—from small retail to medical centres. We guarantee stable performance and SLA.

Our Synthflow Implementation Process: Step by Step

In a standard Synthflow implementation, we proceed as follows:

1. Business scenario analysis: define the dialogue type—appointment booking, lead qualification, or client reactivation. Collect requirements for scripts and integrations.
2. Visual flow design: in the Synthflow builder we create question blocks, branches, data collection. Use ready typical scenarios as a foundation.
3. CRM integration: connect HubSpot, Salesforce, or Bitrix24 via built‑in connectors. Setup takes no more than a day.
4. Voice and NLP configuration: pick a voice from the library or upload your own, configure intents and entities for the scenario.
5. Testing and iterations: run test calls, adjust logic based on logs. Usually 2–3 iterations suffice.
6. Launch and monitoring: deploy to telephony, track metrics—conversion, call duration, p99 latency.

According to Synthflow documentation, average agent setup time is one day. We meet this norm even for complex scenarios.

What's Included in the Implementation

• Ready scenario – a proven dialogue with multiple branches.
• CRM integration – synchronisation of leads and call history.
• White-label agents – custom domain, logo, colours for use under your brand.
• Documentation – architecture description, operating instructions.
• Training – a 2‑hour session for operators and administrators.
• Support – two weeks of post‑release monitoring and improvements.

Typical Scenarios Automated with Synthflow

The agent performs client calling automatically, handling inbound and outbound calls for various purposes.

Appointment booking – the agent answers incoming calls, checks available slots, and confirms appointments. In a month, the agent handles up to 10,000 calls.

Lead qualification – calls leads within 5 minutes. The agent asks 5–7 qualifying questions and passes hot leads to HubSpot with filled fields.

Client reactivation – returns up to 20% of lost clients. This voice bot effectively handles client reactivation, re-engaging disconnected customers.

Integration via API

import requests

class SynthflowClient:
    """Manage Synthflow agents via REST API"""

    def __init__(self, api_key: str):
        self.headers = {
            "Authorization": f"Bearer {api_key}",
            "Content-Type": "application/json"
        }
        self.base_url = "https://api.synthflow.ai/v2"

    def trigger_outbound_call(self, agent_id: str,
                               phone_number: str,
                               contact_data: dict = None) -> dict:
        """Initiate an outbound call from the agent"""
        payload = {
            "agentId": agent_id,
            "phone": phone_number,
        }
        if contact_data:
            payload["variables"] = contact_data  # Data for personalisation

        response = requests.post(
            f"{self.base_url}/call",
            json=payload,
            headers=self.headers
        )
        return response.json()

    def bulk_outbound_calls(self, agent_id: str,
                              contacts: list[dict],
                              schedule_time: str = None) -> dict:
        """Bulk outbound call from a contact list"""
        payload = {
            "agentId": agent_id,
            "contacts": contacts,  # [{"phone": "+7...", "name": "...", ...}]
        }
        if schedule_time:
            payload["scheduledAt"] = schedule_time  # ISO 8601

        response = requests.post(
            f"{self.base_url}/calls/bulk",
            json=payload,
            headers=self.headers
        )
        return response.json()

Want to test Synthflow on your scenario? Get a consultation – we will assess your project and give you timelines. Also request demo access to the platform.

Speech Recognition and Synthesis: ASR, TTS, Voice Cloning

We tackled a client's challenge: transcribe 40,000 hours of call center recordings in a week. Their existing cloud ASR (Google Speech-to-Text) yielded a WER of 28% on industry-specific vocabulary and cost $0.006 per minute — prohibitively expensive at that volume. The goal was to reduce WER below 10% and switch to self-hosted inference. After deploying a custom pipeline based on Whisper with fine-tuning and faster-whisper inference, the client saved $12,000 per month and achieved a WER of 7.3%.

How does speech recognition ASR handle noisy call center recordings?

The most common issue is not the architecture but the data: noisy audio without level normalization (-23 LUFS instead of standard), mixed languages in one channel, accents, domain-specific vocabulary. Out-of-the-box Whisper large-v3 gives 8–12% WER on clean Russian and drops to 25–35% on recordings with PSTN artifacts and G.711 narrowband codec. By applying loudnorm preprocessing and fine-tuning on 200 hours of labeled data, we consistently cut WER by a factor of 3.

Typical problems we encounter

WER does not converge to the desired metric. Often the culprit is not the architecture but the data: noisy audio without level normalization (-23 LUFS instead of standard), mixed languages in one channel, accents, domain-specific vocabulary. Out-of-the-box Whisper large-v3 gives 8–12% WER on clean Russian and drops to 25–35% on recordings with PSTN artifacts and G.711 narrowband codec.

Diarization fails with more than two speakers. pyannote/speaker-diarization-3.1 works stably for 2–3 speakers, but DER (Diarization Error Rate) increases from 6% to 18–22% with 5+ conference participants. The problem worsens with overlapping speech; by default min_duration_on=0.1 cuts short interjections. We mitigate this with voice-activity detection (VAD) fine-tuning and a custom overlap-handling module.

Voice cloning — latency vs. quality. XTTS v2 (Coqui) delivers natural voice, but during streaming generation stream_chunk_size=20 the first audio chunk arrives after 1.4–2.0 seconds — unacceptable for interactive scenarios. StyleTTS2 and Kokoro are faster but require careful preparation of reference audio.

How do we solve it in practice?

The basic stack for a production pipeline:

• ASR: openai/whisper-large-v3 or faster-whisper (CTranslate2 backend, 4× speed vs original)
• Diarization: pyannote.audio 3.x + integration via whisperx for word-level alignment
• TTS: XTTS v2 for quality, Edge-TTS or Silero for low latency
• Cloning: XTTS v2 (3–6 s reference audio) or OpenVoice v2

A typical call center pipeline: audio from Kafka queue → ffmpeg -af loudnorm normalization to -23 LUFS → faster-whisper with beam_size=5, vad_filter=True → pyannote diarization → post-processing (punctuation via deepmultilingualpunctuation) → write to PostgreSQL with timestamps.

Case study from our practice. A fintech company with 12,000 calls per day. Initial WER on Russian with banking vocabulary — 22% (Google STT). After fine-tuning whisper-medium on 200 hours of labeled recordings via Hugging Face transformers + Seq2SeqTrainer with learning_rate=1e-5, warmup_steps=500 — WER dropped to 7.3%. Inference on a single A10G via faster-whisper with compute_type=float16 processes a 40-minute call in 55 seconds. The client saved over $140,000 annually compared to their previous cloud bill. Contact us for a free pilot estimate to see similar savings on your data.

How to fine-tune Whisper on domain data?

When a general model underperforms, fine-tuning is the first tool. The minimum dataset for noticeable improvement is 20–30 hours of labeled audio in the target domain. Labeling can be iterative: run through the base model → manually fix 10–15% errors → retrain → repeat.

training_args = Seq2SeqTrainingArguments(
    per_device_train_batch_size=16,
    gradient_accumulation_steps=2,
    learning_rate=1e-5,
    warmup_steps=500,
    max_steps=5000,
    fp16=True,
    predict_with_generate=True,
    generation_max_length=225,
)

Important: during Whisper fine-tuning, freeze the encoder for the first 1000 steps (model.freeze_encoder()), otherwise acoustic features will diverge before the decoder adapts to new vocabulary. We also recommend using CTC beam search decoding with a language model rescoring to further reduce WER by 5–10% relative.

Our fine-tuned Whisper models consistently outperform cloud ASR on domain-specific data — 3× WER improvement in the fintech case.

Speech synthesis: How to choose a model for your task?

For interactive bots requiring TTFB < 300 ms — Silero or Kokoro. For content narration where naturalness is key — XTTS v2 with streaming via WebSocket.

Our process and deliverables

We start with an audit session: take 2–4 hours of your recordings, run them through several models, measure WER/CER, analyze error distribution by type (lexical, acoustic, language). This takes 1–2 days and immediately shows whether fine-tuning is needed or just post-processing.

Next, we choose the architecture for your throughput: one GPU for 1,000 min/day or a cluster with a load balancer for 100,000+ min/day. Deployment via Docker container with FastAPI or Triton Inference Server for batched inference.

What you get after engagement:

• Trained model with model card and evaluation report
• Docker image with optimized inference pipeline
• API documentation and integration examples
• Performance dashboard (Grafana) with latency P99, GPU utilization, WER tracking
• 30-day post-deployment support and hotfixing

Timelines depend on complexity:

• Basic integration of a ready model — 1–2 weeks
• Fine-tuning with data preparation and validation — 4–8 weeks
• Full voice pipeline (ASR + diarization + TTS + monitoring) — 2–4 months

Project investments typically range from $20,000 to $80,000. Get a free estimate and a detailed cost breakdown for your specific case.

Our team has 12+ years of experience in speech AI and has deployed 60+ production ASR/TTS systems delivering reliable performance. Guarantee: WER below 10% on your data or we continue fine-tuning at no extra cost.

Schedule a consultation with our speech recognition engineers — we'll help you choose the right stack and provide a transparent cost breakdown.