Agente de IA Voz Text to Speech & Speech to Text
Budget: €1,500 – €3,000 EUR
An AI agent needs to be developed that works with open source LLM, can be given a training prompt, and when connected to a virtual VoIP switchboard, can transform speech to text, check the database using a search function and algorithm (already developed), and then respond to the user in real time during the call by transforming text to speech.
- It is very important that the speech is in Spanish. The Spanish voices must be very well done. Examples of voices we like are attached. It should be possible to train the new voice with one of the examples.
- We will ask the selected company for at least 2-3 examples of Spanish voices in advance.
- We do not want to use ElevenLabs, as it will be used for more than 1000 minutes/day of calls. Using ElevenLabs would incur a cost of around €8000/month, which we want to avoid. We are looking to use open source models in most technologies.
-------
Proposed example:
9-hour workday (9:00 a.m. to 6:00 p.m.) Monday through Saturday (24 days)
2,080 minutes of audio generated/transcribed per day.
3,750 minutes of call time (PBX).
Eleven Labs case study
Business Plan: 3,600 min/month included (120 min/day) = $1,320/month
Extra minutes: 1,960 min/day × $0.08/min = $156.80
Elevenlabs monthly total: $3,763.20/month
VAPI (Used with elevenlabs to connect to switchboards)
Minutes per day (3,750) × Rate per minute (0.05) = $187.50/day
Total monthly cost for Vapi: $4,500/month
Final total: $8,263.20/month* (depending on minutes, price is only an example)
* Connection tools such as n8n.io have not been taken into account, which would incur additional resource costs, normally for each execution/call.
- It is very important that the speech is in Spanish. The Spanish voices must be very well done. Examples of voices we like are attached. It should be possible to train the new voice with one of the examples.
- We will ask the selected company for at least 2-3 examples of Spanish voices in advance.
- We do not want to use ElevenLabs, as it will be used for more than 1000 minutes/day of calls. Using ElevenLabs would incur a cost of around €8000/month, which we want to avoid. We are looking to use open source models in most technologies.
-------
Proposed example:
9-hour workday (9:00 a.m. to 6:00 p.m.) Monday through Saturday (24 days)
2,080 minutes of audio generated/transcribed per day.
3,750 minutes of call time (PBX).
Eleven Labs case study
Business Plan: 3,600 min/month included (120 min/day) = $1,320/month
Extra minutes: 1,960 min/day × $0.08/min = $156.80
Elevenlabs monthly total: $3,763.20/month
VAPI (Used with elevenlabs to connect to switchboards)
Minutes per day (3,750) × Rate per minute (0.05) = $187.50/day
Total monthly cost for Vapi: $4,500/month
Final total: $8,263.20/month* (depending on minutes, price is only an example)
* Connection tools such as n8n.io have not been taken into account, which would incur additional resource costs, normally for each execution/call.