We have hands-on experience with speech and voice recognition, and with an AI voice agent that calls customers and talks to them. We do not name our clients – our confidentiality agreements require it.
Included
The recording becomes text with speakers separated and timestamps added, so you can quickly find a moment and listen to it.
AI writes a short summary of each conversation: what was discussed, what was agreed, and who does what by when. We agree the summary format with you.
The text and summary are saved to the customer record in your CRM, emailed or stored in your document system automatically – no copying.
Product names, surnames, abbreviations and industry terms go into a glossary so they are recognised and spelled consistently.
We prepare SRT or VTT subtitle files for YouTube, your website player or a training platform.
In a form on your website or in an internal system, people can speak instead of typing: the text appears in the field and can be corrected before saving.
Only the staff who need recordings and transcripts can see them. We agree the retention period with you so it matches your privacy policy.
We show your team how to review and correct transcripts and extend the glossary. Programming bugs are fixed free of charge for 12 months.
Price
Connecting speech recognition to one recording source and one system costs from €390: this covers transcription with speaker separation, an AI summary, a glossary of terms and delivery to your CRM or email. The speech recognition service (API) is billed separately by recording length. One-off files – meetings, interviews or videos – are quoted by volume.
Price from
€390
Get a quotePrices are final – no VAT is charged (we are not VAT-registered). The final price depends on the project: we fix it and the timeline in the contract before work starts. We do not promise accuracy percentages – accuracy depends on recording quality, so we test with your recordings first.
Timeline
Transcription with summaries connected to one system goes live within 3–7 working days of receiving the deposit and access details. Subtitles or voice input in forms take 2–5 working days. If you start with a pilot, the first 1–2 weeks are spent checking quality on your real recordings.
| Solution | What is included | Timeline |
|---|---|---|
| Subtitles or voice input | subtitle files or a voice input button in a form, glossary | 2–5 working days |
| Calls or meetings into your CRM | text, summary, tasks, delivery to CRM or email | 3–7 working days |
| One-off files | text or subtitles from your recordings | after assessment |
Process
We discuss what recordings you have, where they come from, which languages are spoken and where the text or summary should go.
Using a few of your recordings, we show you what the text and summary look like. This shows straight away whether the audio quality is good enough.
We recommend starting with an AI pilot: for 1–2 weeks, the solution runs on your real conversations, your team checks the results, and then we decide together whether to continue.
We connect the recording source to speech recognition and your CRM or email, and set up the glossary and summary template.
We show your team how to review and correct transcripts, find recordings and extend the glossary.
During the first month, we monitor quality and costs and extend the glossary. Programming bugs are fixed free of charge for 12 months.
Comparison
| Criterion | Manual notes | Generic transcription app | Integrated speech recognition (Saitai.lt) |
|---|---|---|---|
| Who writes it down | an employee during or after the call | the app, but you upload files yourself | automatically after every conversation |
| Where the text ends up | wherever the employee wrote it | in the app account | in your CRM, email or document system |
| Summary and tasks | if the employee writes them | generic format | based on your template |
| Your terminology | known to the employee | often misspelled | glossary of terms |
| Price | staff time | per-user subscription | setup from €390 + API by recording length |
| Best for | a few calls a week | occasional recordings | many conversations whose results must reach your systems |
Examples
These examples show how the solution works. They are not specific client cases – we do not name our clients.
Send us a sample recording – we’ll show you the result
Tell us which conversations or recordings you want turned into text and where it should go. Within 1 working day, we’ll explain how it would work and what it would cost.
FAQ
Current models handle clearly recorded Lithuanian well, but they make mistakes with background noise, dialects, fast speech and rare terms. That is why we test the solution on a few of your recordings before quoting and add your terminology to a glossary. We do not promise accuracy percentages – we show you the result on your own recordings.
Yes, as long as customers are told in advance – for example, at the start of the call – along with the purpose and retention period. GDPR requires this. We recommend updating your privacy policy and informing staff about recording. We can help draft the notice, but we do not provide legal advice.
We test with your recordings first. A different microphone or phone system settings, noise reduction before recognition and a glossary often help. If the recordings are still poor, we tell you straight away and suggest human review.
Recordings and transcripts are stored in your system or on our server in the European Union, and we agree the retention period with you. For recognition, audio is sent to the speech recognition provider via the API – we check their terms in every project, including whether data is used to train models.
We deliver subtitles in SRT or VTT format, which work with YouTube, Vimeo and most website players and training platforms. You can also get the text as a DOCX or plain text file.
Yes, we have experience with an AI voice agent that calls customers and talks to them: it answers simple questions, books or confirms appointments and passes complex cases to a member of staff. The agent introduces itself as AI. We quote the price and timeline separately.
More
The other way round: articles, products and videos with an AI voice in Lithuanian – from €290.
When AI needs not only to transcribe a call but also to take action in your CRM – from €1,490.
Where call summaries and tasks are linked to your customers – from €1,900.
Prices for all our AI services in one place, from the free assessment to maintenance.
Enquiry
Tell us what recordings you have (calls, meetings, videos), which languages are spoken and where the text should go. Tel. +370 670 99265. We work in English by email, chat and messengers; calls by arrangement. We also speak Lithuanian and Russian.
+370 670 99265Mon–Fri 9:00–18:00
info@saitai.ltWe reply within 1 working day
Alaušo g. 1C, LT-08424 VilniusMeetings by prior arrangement by phone only
